by
Samuel B. Kesner
BS Mechanical Engineering
Massachusetts Institute of Technology, 2006
Submitted to the Department of Mechanical Engineering in Partial Fulfillment of the Requirements for the Degree of
Masters of Science in Mechanical Engineering at the
Massachusetts Institute of Technology
September 2007
© 2007 Massachusetts Institute of Technology
All Rights Reserved
Signature of Author.
Certified by ..........
Department o1 Mechanical Engineering
August 10, 2007
Steven Dubowsky
ProfessY of Mechanical Engineering
Thesis Supervisor
Professor Lallit Anand
Chairman, Committee on Graduate Studies
Accepted by ............
MAssACHUSErS INS
OF TEIHNOLOay
JAN O
by
Samuel Kesner
Submitted to the Department of Mechanical Engineering on August 10, 2007, in partial fulfillment of the requirements for the degree of Masters of Science in
Mechanical Engineering
ABSTRACT
Small hopping robots have been proposed that offer the potential to greatly increase the reach of unmanned space exploration. Using hopping, bouncing, and rolling, a small spherical robot could access and explore subterranean areas, such as craters and caves, on distant planets. Hopping mobility allows the robot to overcome larger obstacles than conventional wheeled rovers. Bouncing and rolling allows the robot to infiltrate underground areas too challenging and dangerous for manned exploration. The robots would use onboard sensors to explore and search for signs of water, biological material, and other items of interest to scientists.
This thesis studies the power and mobility feasibility of the Microbot hopping robot concept. One of the most important mobility issues for autonomous robots is the availability of energy and how that energy is used. The Microbot utilizes a hydrogen fuel cell power system. A fuel cell power system design is proposed and an experimental prototype device was constructed and tested. The results presented indicate that a miniature hydrogen fuel cell power system is a feasible energy generation option for the
Microbot system concept.
The feasibility of the hopping mobility system is also investigated. An integrated power consumption model of the Microbot is proposed and the ability of the Microbot power and mobility systems to complete a Martian reference mission is demonstrated.
Simulated studies of the mobility system's capacity to overcome obstacles and navigate the Martian terrain are presented. The results of these simulations are analyzed and the mobility and power system design tradeoffs are examined. Finally, recommendations for future research are made.
Thesis Supervisor: Steven Dubowsky
Title: Professor of Mechanical Engineering
3
This research was conducted in the MIT Field and Space Robotics lab with funding from the NASA Institute for Advanced Concepts
I would like to thank the many people who made this thesis possible:
" Dr. Dubowsky for giving me the opportunity to learn and working with me on my writing and analytical thinking.
* Dr. J.S. Plante for his endless and continued advice, guidance, criticism and mentoring.
* Dr. Karl lagnemma for helping with the terramechanics work.
* Dr. Tibor Fabian for providing the fuel cells and technical expertise that made my experimental work possible.
" Steve Peters for helping me with ADAMS, especially when it was most frustrating.
" Peggy Boning for being my peer and advisor.
" The entire FSRL for helping me with every challenge I have encountered.
" My family for supporting my learning over the past 24 years with love and guidance and making sure that I never work too hard.
" The lovely Ms. Jess for providing more emotional support and love than I could ever hope for.
And finally, I would like to thank MIT for educating me for the last 5 years. I owe my development to you.
4
ABSTRACT ....................................................................... .
------------------------- 3
ACKNOWLEDGEMENTS..........................................................................
4
TABLE OF CONTENTS ............................................................................
..... 5
LIST OF FIGURES ......................................................................
LIST OF TABLES ...................................................................
.. 7
.
.........
CHAPTER 1: INTRODUCTION...........................................................................
1.1 M otivation.......................................................................................
10
........... 11
1.2 Background Literature ....................................................................................
12
1.2.1 Hopping Robots .................................................................................... 13
1.3
1.2.2 Fuel Cell Power Systems for M obile Robots........................................ 15
M icrobot Concept ........................................................................................... 16
1.3.1 M icrobot Subsystems Overview ...........................................................
17
1.3.2 M obility System .....................................................................................
18
1.3.3 Power System.......................................................................................... 19
1.3.4 Sensor Payload....................................................................................... 19
1.3.5 Electronics.............................................................................................. 20
1.4 Research Overview ......................................................................................... 21
CHAPTER 2: FUEL CELL POWER SYSTEM .....................................................
23
2.1 Fuel Cell Power Generation........................................................................... 23
2.1.1 Fuel Cell Operation................................................................................ 24
2.1.2 Implementation for M obile Robots........................................................ 27
2.2 Experimental W ork......................................................................................... 29
2.2.1 Design and Construction...................................................................... 29
2.2.2 Experimental Results and Analysis ..................................................... 33
CHAPTER 3: MICROBOT-TERRAIN INTERACTIONS ......................................
3.1 M icrobot-Terrain Interaction Physics .............................................................
3.1.1 Terramechanics....................................................................................
3.1.2 Hopping Efficiency and Terrain Energy Dissipation............................
3.2 Terrain Interaction Experiments ....................................................................
3.2.1 Experimental Setup................................................................................
3.2.2 Experimental Results .............................................................................
41
41
42
46
48
48
50
CHAPTER 4: FEASIBILITY ANALYSIS ............................................................
53
4.1 System Level Feasibility.................................................................................... 53
4.1.1 System M odel ........................................................................................... 54
5
4.1.2 System Efficiencies...................................................................................
4.1.3 System Analysis Results ...........................................................................
4.2 Application Feasibility....................................................................................
4.2.1 Simulation Development ......................................................................
4.2.2 Overcoming Obstacles ...........................................................................
4.2.3 Reference Mission Simulations .............................................................
4.3 Design Tradeoffs............................................................................................
CHAPTER 5: CONCLUSIONS AND RECOMMENDATIONS .................................
79
5.1 Summary of Results........................................................................................ 79
5.2 Future W ork................................................................................................... 80
5.3 O utlook .............................................................................................................. 8 1
REFERENCES .............................................................................................
82
69
73
77
56
59
64
64
APPENDIX
A: DETAILS OF THE FUEL CELL EXPERIMENT ............................. 87
APPENDIX
M OBILITY SIMULATION DETAILS ..........................................
90
APPENDIX C: GENERATING THE SIMULATION TERRAIN ...............................
6
LIST
Figure 1: An orbiter view of a lava tube cave with skylights on Mars [4]. ...................... 11
Figure 2: An artist's representation of (a) the Microbot concept and (b) its progression through rubble [13]. ....................................................................... 12
Figure 3: The Microbot system concept and major modules [13]............................... 17
Figure 4: The FSRL experimental Microbot dielectric elastomer actuator [28]. .......... 18
Figure 5: (a) A single chip Mote developed at UC Berkeley [55]. (b) The off-theshelf component required to regulate the power system [27]............................ 20
Figure 6: A diagram of how a PEM fuel cell converts hydrogen and oxygen into electricity, w ater, and heat [52]. ....................................................................... 24
Figure 7: A typical PEM fuel cell voltage-current curve [2]................... 26
Figure 8: The three air-breathing PEM fuel cells and the gas humidification b eak er..................................................................................................................... 30
Figure 9: A schematic of the fuel cell power system prototype. ................................. 31
Figure 10: The experimental fuel cell power system................................................... 32
Figure 11: The fuel cell power output (a) without charging the Li-ion battery and
(b) w ith charging................................................................................................ 35
Figure 12: The DEA power input during two actuations............................................... 36
Figure 13: A diagram of the Microbot-terrain interaction analysis. ............................. 43
Figure 14: A Opportunity rover wheel stuck in a soft Martian sand pile [37].............. 44
Figure 15: Two views of the impact indents made by the Microbot mockup in flour: (a) Top view of the indent with a ruler and indent outline (b) an orthographic view of the indent......................................................................... 51
Figure 16: A concept drawing of Microbots entering a cave on Mars [13].................. 54
Figure 17: The Microbot energy conversion model. .................................................... 55
Figure 18: A m initurePEM fuel cell [1]............................................................................ 57
Figure 19: A miniature high voltage DC-DC converter [15]....................................... 57
Figure 20: A current prototype of the Microbot system capable of hopping. It contains a DEA actuator, energy storage, and integrated electronics [27]........ 58
Figure 21: A plot of the total number of hops on Mars during a 7 Earth day mission as a function of the Microbot mass. .................................................... 62
Figure 22: The percent of the fuel that is consumed by the mobility system during a 7 Earth day m ission to M ars. ......................................................................... 63
7
Figure 23: A diagram of the simulation terrain. ........................................................... 65
Figure 24: A lava tube skylight in Idaho, USA [22]...................................................... 66
Figure 25: The lava tube cave and skylights in the simulated Mars terrain. ................ 66
Figure 26: A Microbot traversing the simulated rock pile............................................ 69
Figure 27: The success rates of all of the simulations as a function hop height........... 71
Figure 28: The rate of entrapment as a function of Microbot diameter and hop h eigh t .....................................................................................................................
Figure 29: The rate of bouncing off as a function of Microbot diameter and hop h eigh t. ....................................................................................................................
72
72
Figure 30: The predicted and simulated path of Microbots with five different num bers of hop directions. ................................................................................. 75
Figure 31: A diagram of two simulated Microbot trajectories over the surface, into the skylight, and down the lava tube cave: (a) a traverse over the hill and
(b) a traverse around the hill............................................................................. 76
Figure 32: Microbots hopping over rock piles in a lava tube simulation. ..................... 77
Figure 33: The setup used to measure the bubble volume and approximate the fuel flow rate..................................................................................................... . . 89
Figure 34: The ADAMS simulation Simulink control system. .................................... 92
8
LIST
Table 1: The Microbot concept system specifications................................................. 17
Table 2: A comparison of current hydrogen storage methods [38]. ............................. 27
Table 3: The experimentally measured power values for the system........................... 35
Table 4: The power system efficiencies......................................................................... 37
Table 5: The averaged fuel cell prototype performance values and the predicted
Microbot power system performance specifications from section 4.1............. 38
Table 6: The terramechanics analysis parameters used in this analysis [26]................ 45
Table 7: Microbot sinkage values predicted by the terramechanics analysis............... 46
Table 8: The results of the Microbot-terrain interaction experiments in sand.............. 51
Table 9: The Microbot subsystem efficiencies used in the system level study. ........... 59
Table 10: The Microbot design values determined by the system level study.............. 63
Table 11: The impact contact model values. ................................................................ 68
Table 12: The rock size distribution in the rock pile obstacle....................................... 70
Table 13: The M icrobot model parameters.................................................................... 91
9
CHAPTER
1
This thesis reports on the power and mobility feasibility of the previously published Microbot space exploration concept [13]. The Microbot concept is proposed to explore the extraterrestrial (ET) caves of the solar system with hundreds or thousands of small robots. The Microbots would hop, bounce, and roll across the surface of the planet and into subsurface features of interest. The robots that successfully enter the features would use their onboard sensor payload to take scientific measurements [27].
Planetary scientists are interested in ET caves because they may contain water deposits, geological information, or even possibly biological material
[5]. Subterranean
ET exploration can be exceedingly dangerous and therefore is a strong candidate for robotic exploration. Current wheeled space robot designs, such as the Mars Exploration
Rover (MER) and the Mars Science Laboratory (MSL), are not well suited for the large obstacles and steep slopes of subsurface exploration [13]. A rough terrain entrapment in an ET cave would cause a single point mission failure, thus ending a single-robot mission.
This research demonstrates the feasibility of certain aspects of the Microbot concept, specifically the fuel cell power system and the mobility concept. A hydrogen fuel cell power system design is proposed, constructed, and tested. The robot-terrain interaction is examined and performance predictions are made. An analysis of the power and mobility systems demonstrates the feasibility of the concept for a reference mission
Chapter 1: Introduction
10
to Mars. Finally, simulation results of the hopping mobility system are presented and design tradeoffs are considered.
The research was conducted in the MIT Field and Space Robotics Lab (FSRL) under Dr. Dubowsky with funding from the NASA Institute for Advanced
Concepts
(NIAC), an agency that supports research projects that have the potential to significantly impact the aeronautic and astronautic fields in 10 to 40 years [44].
1.1 Motivation
There is a significant scientific interested in exploring ET subterranean areas, including caves and lava tubes on the Moon and Mars (see Figure 1). Astrobiologists and geomorphologists are interested in these areas because they may contain water deposits, geomorphologic information, or possibly biological material. Scientist are interested in
ET caves and lava tubes because they provide shelter from the harsh environmental conditions on the surface, hence sensitive materials have a greater chance of being located and sampled in those locations [4, 5]. As shown in Figure 1, access to these ET features is provided through collapsed roof sections called skylights.
Figure 1: An orbiter view of a lava tube cave with skylights on Mars [4].
The current challenge preventing missions to these locations is the lack of a suitable exploration platform. Astronaut exploration is not possible due to the high risk
I I
Chapter 1: Introduction
associated with penetrating possibly kilometers into an uncharted cave composed of steep inclines and rock outcroppings. Current planetary rovers are also unsuited for subsurface exploration because they can not overcome the large obstacles present in a partiallycollapsed lava tube or a cave opening. The solution to this technological need is the
Microbot concept, discussed in greater detail in section 1.3. Figure 2 shows an artist's representation of the concept.
The purpose of this thesis is to determine the power and mobility feasibility of the
Microbot concept design. A mass-energy analysis of the system level design is conducted. A design for an experimental fuel cell power system is proposed, tested, and analyzed. Finally, the feasibility of the hopping, rolling, and bouncing mobility system is examined through simulations and the performance is analyzed.
(a) (b)
Figure 2: An artist's representation of (a) the Microbot concept and (b) its progression through rubble [13].
1.2 Background Literature
This section presents a brief review of the literature on hopping robot mobility systems and the application of fuel cells to mobile robots. The feasibility of the Microbot
Chapter 1: Introduction
12
concept is highly dependant on the successful performance of the mobility and fuel cell power systems.
1.2.1 Hopping Robots
A great deal of research has been conducted on hopping robots. Hopping mobility systems offer a number of distinct advantages over rolling or walking mobility approaches. Although the maximum sized obstacle a wheeled robot can clear is limited
by its wheel diameter and traction force, a hopping robot can clear objects many times its size and is only limited by its hopping height. This quality is advantageous when traversing rough and obstacle-filled terrain. Unlike articulated wheeled robots, hopping robots do not have to adjust their configuration to overcome larger rocks or uneven terrain. Hopping mechanisms also require less actuators and linkages than equivalent walking or rolling mechanisms. This attribute results in a decrease in the mobility system complexity and weight over other mobility mechanisms, a significant advantage for space exploration missions [19].
Hopping robot mobility systems can be divided into two major categories: continuous and discontinuous hopping systems. Continuous hopping is the dynamically stabilizing approach where the robot is in continuous oscillatory motion. The biomimetic equivalent for this hopping motion is a kangaroo. The best known research in this field is the work of Marc Reibert at the MIT Leg lab [43]. Reibert's robots are marked by their energy-regenerative legs and active balancing. Though continuous hopping robots can perform impressive feats, they consume a great deal of energy and are not robust in unstructured environments [43].
The other major hopping approach, discontinuous hopping, uses a "hop and wait" method favored by hopping insects and reptiles. The first documented application of discontinuous hopping to autonomous robotics is the soviet PROP-F robot designed to
Chapter 1: Introduction 13
explore the low gravity moon of Phobos in the 1980's [17]. This device used an active hopping foot and stabilizing actuators to move in the low gravity environment. However, it never completed a successful mission. Hopping is advantageous in low gravity because it allows the robot to clear large obstacles than wheeled robots [50].
Discontinuous hopping approaches can be further classified by the hopping actuation technology employed. These technologies include mechanical, electrical, and chemical energy storage mechanisms. Mechanical systems commonly use coil or leaf springs to generate the hopping force. Robots that use this type of mechanism include the
JPL Hopper, the University of Minnesota Scout robot, and the Microbot system discussed in this paper [19, 48, 27]. Mechanical energy storage elements offer good energy storage densities, do not produce much heat, are highly repeatable and efficient, and allow for a large range of design configurations. These factors make springs and bistable elements the most commonly selected hopping energy storage devices for discontinuous hopping mobility systems.
Purely electrical systems, including linear motors and magnetic systems, do not offer large enough power densities or forces in a mobile package. Thus, fully autonomous electric hopping mobility systems are not possible. Researchers have selected these technologies when large hopping forces are not required, such as in microgravity environments. The best known example of hopping microgravity robotics is the asteroid exploration MUSES-C robot from the University of Tokyo, later renamed the Hayabusa project [45,46]. Research considered a number of actuation methods for this robot, including linear actuators and magnetic levitation systems. Unlike other hopping mechanisms, the Hayabusa robot hops by displacing an internal mass with enough force to move the robot in a microgravity environment.
The final approach to hopping mobility is chemical fuel systems. These mechanisms involve the combustion of some fuel or explosive. The appeal of
Chapter 1: Introduction 14
combustion is that high explosives and hydrocarbon fuels offer a larger energy storage density than any battery technology. Also, chemical energy can be quickly released, thus generating the hopping motion without complex mechanisms or transmissions. The best know example of these devices is the Sandia National Lab hopper. It uses a piston and fuel ignition system similar to an internal combustion engine to generate a hop [16].
While chemical fuel combustion systems do offer large energy and power densities, they are not appropriate for space robotics because they pollute the environment and therefore may affect or damage the sensors and scientific samples.
1.2.2 Fuel Cell Power Systems for Mobile Robots
Though fuel cell research has been conducted for over sixty years, limited experimental work has been published on the application of fuel cells to small mobile space robots [30]. Fuel cells have been used in space exploration since the Gemani V spacecraft in 1965 [21,8]. Fuel cells provide electricity, heat, and drinking water to spacecraft crews.
All autonomous exploration robots to date use batteries, solar panels, radioisotope thermoelectric generators (RTGs), or a combination to generate power. NASA and ESA have conducted a number of studies exploring the possibility of using fuel cells on future planetary missions [21,8]. However, fuel cells have yet to be included in any unmanned space exploration missions because current miniature fuel cell technology does not outperform batteries [7].
The most relevant research to date on this topic is the work done by Sandia
National Labs on integrating a fuel cell system into its Robotic All Terrain Lunar
Exploration Rover (RATLER) [58]. The RATLER is a medium-size surveillance and exploration robot with an articulated tank-track mobility system. Researchers installed an experimental PEM fuel cell system on the robot to extend the vehicle's range [58].
Chapter 1: Introduction
15
The other relevant research in this field is the work by researchers at Queen's
University in Ontario, Canada to adapt a micro-fuel cell power system for a mobile robot
[56]. This work discussed the technical issues required to adapt an off-the-shelf airbreathing hydrogen fuel cell for a mobile application. The Queen's University paper illustrates how challenging it currently is to construct a stable and portable miniature fuel cell system due to thermal regulation, hydrogen storage, and water management issues.
Although the researchers were able to power the robot with a hydrogen fuel cell system, the currently available fuel storage and the fuel cell regulation equipment size and weight must be decreased to allow for a fuel cell system to outperform current battery technology [56]. Therefore, this thesis determines what specific power system components must be improved for the current technology to meet the Microbot system specification requirements.
1.3 Microbot Concept
This section summarizes the previously published work on Microbot concept
[12,13,27]. The Microbot system is an exploration robot designed to traverse very rough terrain and explore ET subterranean features of interest while being robust to single point failures [13]. Upon reaching a feature, the Microbots will use their hopping, bouncing, and rolling mobility system to descend into and explore the feature's interior while taking scientific measurements, a task currently infeasible with today's rover technology. The
Microbots would then transmit the data to a surface lander or orbiting spacecraft via a low power local-area communication network established between the members of the team [12,13]. Figure 3 shows a diagram of the Microbot concept.
Chapter 1: Introduction 16
Transparent protective spherical shell
(Polycarbonate]
360 deg camera
Indexing EPAM actuator
7(Shown contracted)
(2x] fuel tanks
Detection sensors
(sniffers, heat, organic...)
Electronics (Computation, communication) and power generation (fuel cell(
-
Bistoble energy oa device
(shown extended)
Figure 3: The Microbot system concept and major modules [13].
The individual Microbots will work collaboratively to navigate, take scientific measurements, and transmit the data to the surface. This
highly redundant approach to
ET exploration allows the overall robotic system to be robust to failure or entrapment of any number of the team members. Table 1 summarizes the Microbot system design specifications [27].
Table 1: The Microbot concept system specifications.
Parameter
Microbot Diameter
H o .H eight -{ *-'--'--- "-----
Microbot Mass
Mi._tn
Max. hop frequency
Values
100 mm
I m ...
__
2000 hop
2 hops / minute
1.3.1 Microbot Subsystems Overview
The individual Microbots consist of four major subsystems: a mobility system, a fuel cell power generation system, an integrated electronics package, and a scientific sensor payload [12].
17
17
1.3.2 Mobility System
The main component of the Microbot mobility system is a dielectric elastomer actuator (DEA) coupled with a bistable device (see Figure
3). The DEA works by using the Maxwell (electrostatic) pressure generated by a strong electric field to compress a soft elastomeric film, thus generating an expansion in the directions orthogonal to the film
[39]. The current state-of-the-art actuators allow for the film area to expand up to several times its initial size during actuation [40, 54]. Figure 4 shows the experimental Microbot
DEA device, power transmissions, and energy storage springs. The current DEA prototype uses springs instead of a bistable mechanism.
Cone DEA
-4 Springs
For the Microbot concept to initiate a hop, electrical energy is slowly pumped into the DEA, causing it to expand a bistable device. When the bistable mechanism has been deformed sufficiently, it switches states and quickly releases its stored mechanical energy in the form of a hop
[27]. The peak power requirement from the fuel cells is decreased because the energy required to hop the Microbot is accumulated over a relatively long period of time. A proof-of-concept of this bistable hopping mechanism has been demonstrated [40].
The Microbot travels by hopping followed by bouncing and rolling [12]. The hopping direction is controlled by the angle of the bistable device. The Microbot is
18
Chapter 1: Introduction
weighted so that after the rolling and bouncing motion is complete, it will return to an upright position [13]. If the environment prevents self-righting, the Microbot will use its actuator to right itself.
1.3.3 Power System
The Microbot system uses miniature hydrogen fuel cells to generate power. Fuel cells were selected as the power source because they outperform battery power supplies in terms of energy density (energy per unit mass) for long duration missions [6, 13].
Fuel cells were also selected because they are high-efficiency but low-power devices.
They will perform well with the low peak power characteristic of the mobility system's
DEA actuators [13].
Proton exchange membrane (PEM) hydrogen fuel cells were selected for the
Microbot system. These fuel cells convert the chemical energy of hydrogen and oxygen into electrical energy, water, and heat. The hydrogen is stored in a metal hydride container and the oxygen in a compressed gas vessel. PEM fuel cells offer the advantages of being very efficient and operating at low temperatures [38]. The small water byproduct can be captured and held within the Microbot to avoid contaminating the environment.
1.3.4 Sensor Payload
The Microbot requires sensors to navigate and make mission specific scientific measurements. To control its mobility systems and navigate through the environment, the Microbot will use sensors including accelerometers, an inertial measurement unit
(IMU), and a vision system [12].
The scientific sensor suite will be selected to meet the mission objectives.
Possible sensors include microscopes, panoramic cameras, mass spectrometers, gas
Chapter 1: Introduction
19
analyzers, chemical sensors, and X or Alpha-Ray sensors. The sensor selection will take advantage of the large number of Microbots in a team by allocating different sensors to each robot, resulting in a diverse range of scientific measurements [12,13].
1.3.5 Electronics
The Microbot concept requires a number of different electronics subsystems, including communication, computation and data storage, and power regulation
[12]. The
Microbots need to communicate with each other and to transmit scientific data to a landing or orbiting craft. High-frequency radio communication will allow for low power communication over long distances on the surface as well as small transmitter and receiver size [32]. This type of ultra-low power and ultra-compact communication technology has been developed, including the single chip Mote shown in Figure
5(a) that can send and receive wireless communication with minimal external components
[55].
Communication from a subsurface location would be possible with the use of a local area network (LAN) where each of the Microbots operate as a network node. The
Microbots could be instructed to stop at various penetration depths in order to maintain a communication link with the surface [12,13].
Non-line-of- sight communication has been shown to be possible in terrestrial caves at distances up to 20 meters
[5].
(a)
(b)
Figure 5: (a) A single chip Mote developed at UC Berkeley [55]. (b) shelf component required to regulate the power system [27].
The off-the-
Chapter 1: Introduction
20
The Microbots also require data processing, data storage and power regulation electronics [12]. The current state of electronics in these areas is close to meeting the requirements of the Microbot system. Miniaturized computer systems can process Mbps of data in a very small volume at low power [32]. Highly efficient miniaturized electronics are available to regulate the power generated by the fuel cell system, see
Figure 5(b).
The primary objective of this research is to examine the power and mobility feasibility of the hydrogen fuel cell powered Microbot concept. The fuel cell power system and mobility approach are analyzed separately and then the feasibility of the power and mobility system level design is demonstrated. The secondary objective is to investigate how different design tradeoffs affect the performance of the Microbot system and to propose development goals and a timeframe for the completion of the Microbot system.
Chapter 2 focuses on the fuel cell power system. This chapter briefly summarizes the operation of fuel cells and discusses the requirements of the Microbot power system.
An experimental fuel cell power system prototype is presented and the system performance results are analyzed and compared to the predicted performance of the
Microbot concept power system. The results are evaluated in the context of the Microbot system feasibility.
Chapter 3 briefly discusses the mechanics and physics of the Microbot-terrain interactions. This chapter presents an analysis of the robot-terrain terramechanics and calculations of the Microbot sinkage into the deformable soil and energy losses due to the terrain. The analyses are validated with laboratory experiments and the results are used to design the Microbot simulations and conduct feasibility analyses.
Chapter 1: Introduction 21
Chapter 4 presents a feasibility analysis of the integrated Microbot power and mobility systems. The first analysis uses a mass-energy model to determine the number of hops possible during a Mars mission given the performance and power demands of the
Microbot subsystems. This chapter also examines the feasibility of the hopping mobility system through a number of simulations of mission-specific tasks.
This thesis demonstrates the power and mobility feasibility of the Microbot concept. A power and mobility system level study determines that the fuel cell system concept can generate enough energy to power the Microbot for a Mars reference mission.
The results of a fuel cell system prototype experiment validate the power system performance values selected for the system level study. The mobility system is shown to be able to clear obstacles and complete a simulated Martian reference mission of over
2000 hops.
Assuming the rate of current research innovations continues, the fuel cell and
DEA technology should be at an appropriate level for the Microbot system in the next 10 years. With sufficient research funding, the entire Microbot system should be ready for deployment in approximately 20 years.
22
22
CHAPTER
2
This chapter presents the fuel cell research conducted for this thesis. Fuel cell technology and operation is briefly reviewed and the design of a fuel cell system for a mobile robot is discussed. This chapter also presents a design for a Microbot fuel cell power system prototype and experimental results. The prototype system performance is compared to the predicted requirements of the Microbot power system. Based on these results, the feasibility of a hydrogen fuel cell power system for the Microbot concept is demonstrated.
2.1 Fuel Cell Power Generation
Fuel cells were selected for the Microbot concept because they out perform batteries for long mission. This is because unlike batteries, to increase the total energy a fuel cell power source can provide only additional fuel needs to be supplied [7, 13].
While the entire mass and volume of a battery scale with the total energy, only the mass and volume of the fuel storage elements are affected for a fuel cell system. Thus they have a high energy density for long duration missions. Hydrogen fuel cells are also high efficiency energy conversion devices, with efficiencies up to 70% depending on operating conditions.
23
23
2.1.1 Fuel Cell Operation
Fuel cells convert chemical energy into electrical energy. Energy is produced by the reaction of a fuel and an oxidizer in the presence of an electrolyte. In most fuel cells, electricity is produced by stripping electrons from the fuel with a platinum catalyst on the anode side of the cell, sending the electrons through a circuit to do electrical work. The positively-charged fuel is then sent through the electrolyte and then reacts with the oxidizer on the cathode side to produce the reaction byproducts and heat
[2].
A number of different possible fuels and electrolytes exist. Polymer exchange membrane (PEM) fuel cells were selected for the Microbot concept [13].
PEM fuel cells were chosen because of their high energy conversion efficiencies (up to 70%) and low operating temperature of 30-100' C [30]. PEM fuel cells use hydrogen gas as the fuel, oxygen gas or air as the oxidizer, and a special class of proton absorbing polymers as the electrolyte. The hydrogen to oxygen mass ratio for this process is 1:8.
Figure 6 shows a diagram of the PEM fuel cell operation.
Fuel
PEM FUEL CELL
Elecmrical Curmet hoat out
Fusl
In
Aod
E
Air
Catho~d* in
Figure 6: A diagram of how a PEMfuel cell converts hydrogen and oxygen into electricity, water, and heat [52].
System
24
24
Each individual fuel cell is composed of a series of layers, including gas flow channels, gas diffusion layers, a catalyst layer to help dislocate the electrons from the hydrogen, and the PEM electrolyte itself. PEM fuel cells are designed to be thin and flat to maximize the electrolyte surface area. The greater surface area allows for a greater amount of hydrogen to pass through the electrolyte and release its electrons, thus generating a greater current. The voltage of each fuel cell is determined by the energy produced during the oxidation reaction, which is a function of the Gibbs free energy equation [30]. For compactness, a number of these individual cells are often arranged in a larger stack.
The maximum voltage, also known as the open circuit voltage, of a hydrogen fuel cell is calculated from the Gibbs free energy equation: y -g
(2.1) where Agf is the Gibbs free energy value for hydrogen and oxygen, approximately -240 kJ at room temperature, and F is the Faraday constant, 96,485 coulomb/mole. This equation gives a theoretical open circuit voltage of approximately 1.2V [2].
Fuel cells can not generate their predicted open circuit voltage due to internal losses. As the electric load experienced by the fuel cell decreases, the current provided increases. As the current generated by the cell increases, the voltage output decreases due to internal losses. These losses include the activation losses associated separating the electrons from the hydrogen, ohmic losses from the electrical resistances in the cell electrodes and electrolyte, and mass transport losses caused by the flow of the reactants inside the cell [30]. Figure 7 shows a typical performance curve for a PEM fuel cell.
Note that the current is per unit area because the current output of a fuel cell scales with the membrane area.
Chapter 2: Fuel Cell Power System
25
Theoretical open circuit voltage
1-
0 .8
4"0.6'
0.4
U
0.2'
0 0.5 1 1.5
Current Density (A/cm2)
2
Figure 7: A typical PEMfuel cell voltage-current curve [2].
The performance of fuel cells is a function of many factors, including the amount of water present in the device, the temperature of the cell, and the operating pressuring of the reactants. Water is required because proton conduction is directly proportional to the water content in the membrane [30]. However too much water can flood the cell, preventing the reactants from reaching the electrolyte. Another water management challenge is that water is also a byproduct of the hydrogen-oxygen reaction. The common solution to this issue is to humidify the input gases and purge the excess water, thus balancing the cell water content.
The cell also needs to be kept at the right temperature so that the reaction can take place: not so hot that the water will vaporize or so cold that it will freeze. Finally, the operating pressures of the reactants are often elevated to increase the power output.
The pressures must be controlled to prevent any damage to the thin and fragile membrane
[30].
Chapter 2: Fuel Cell Power System
26
2.1.2 Implementation for Mobile Robots
Unlike battery technology, fuel cells require additional components to produce electricity. The minimum components required are fuel and a fuel storage device, an oxidizer and an oxidizer storage device, and a regulation and control system.
Hydrogen can be stored as a compressed gas, as a cryogenic liquid, absorbed into another material like a metal hydride, or as a part of a chemical, such as methanol. The effectiveness of a storage device is measured by the mass storage efficiency, the percent mass of the device that is hydrogen, and the volumetric storage efficiency, the number of grams of hydrogen per liter of storage vessel. Table 2 presents a comparison of these hydrogen storage options.
Table 2: A comparison of current hydrogen storage methods [38].
Storage Method
SPressurized Gas (30 ba)
Pressurized Gas (700 bar)
Ciygenic Lgid
Metal Hydride and
Containment Components
Mass Storage Eff. (% wt.) Volumetric Storage Eff. (g/L)
3.1
4.8
14.
14
33
43
2.0 28
The best storage efficiencies are currently reached with liquid hydrogen.
However, liquid hydrogen must be kept below 200 Kelvin to stay a liquid. This extreme cold can not be maintained in a 100 mm diameter robot. Chemical storage, although stable and often in a liquid state, is also infeasible because it requires additional components to extract and purify the hydrogen and store the chemical byproducts.
Therefore, the only two viable options for hydrogen storage are as a compressed gas or in a metal hydride.
Hydrogen gas has the lowest atomic weight of any molecule at 2 grams/mol.
Therefore, extremely high pressures are required when storing it at a reasonable density
Chapter 2: Fuel Cell Power System 27
as a gas in a cylinder. The compressed hydrogen options presented in Table 2 require that the gas be held at pressures approximately 300 and 700 times standard atmospheric pressure. Such high pressure systems are impractical for the Microbot concept.
The metal hydride option has a number of features that make it the most appealing hydrogen storage option. Metal hydrides act as a sponge that soaks up hydrogen gas at the molecular level at relatively low pressures. Hydrides can absorb large quantities of hydrogen at better volumetric efficiencies than liquid hydrogen. However metal hydrides require additional components, including valves and external coatings, to contain the hydrogen inside the material. These components, depending on the storage material and pressure, can be heavy and bulky, thus decreasing the total mass and volume storage efficiencies [38]. Hydrides are also an appealing technology because they are an area of much research and many promising materials are currently being developed.
Oxygen can be stored as a compressed gas because it has an atomic weight of 16 grams/mole, considerably larger than the atomic weight of hydrogen. Carbon fiber reinforced pressure vessels exist that can withstand pressures of up to 10,000 psi [53].
With a safety factor of two, the Microbot could carry the amount of oxygen needed for the Mars reference mission presented in section 4.1 in an approximately 20 ml container, less than 10% of the total Microbot volume. Thus, a high-strength carbon fiber containment vessel can provide sufficiently light-weight storage of the oxidizer.
A fuel cell power system also requires a regulation and control system to operate.
Fuel cells are electrochemical processors that must be monitored and adjusted to perform optimally. The temperature, power output, and internal pressure should be measured, and the reactant gas pressure and flow rate should be adjusted accordingly [38]. Monitoring the power demands on the robot can also help optimize the fuel utilization. These components can be designed specifically for mobile robot applications to minimize volume, weight, and energy consumption.
Chapter 2: Fuel Cell Power System
28
If a robot requires short, intermittent spikes of power, a capacitor or battery can be used to meet the peak power demands. This is the case for the Microbot, where the hopping mobility system requires a peak of power spaced by minutes of inactivity to allow for scientific measurements and to prevent damage to the DEA actuator [6]. In this way, the fuel cell system can operate like a trickle charging device. Trickle charging allows for the Microbot's fuel cell system to run constantly at its optimal operation point, thus maximizing its efficiency. For the Microbot application, trickle charging allows for a smaller fuel cell system to be selected, thus increasing the power system mass and volume efficiencies.
2.2
A hydrogen fuel cell power system has been proposed for the Microbot concept
[12, 13]. To validate this proposed technology and evaluate its feasibility, a fuel cell power system prototype has been designed and constructed. The system is able to continuously charge a lithium-ion battery, power control electronics, and actuate a cone
DEA similar to the one selected for the Microbot prototype (see Figure 4).
This section introduces the experimental setup and how it operates. Experimental results are presented and used to evaluate the power system feasibility. Refer to
Appendix A for technical details of the experimental device.
2.2.1 Design and Construction
The goals of the fuel cell power system experiment are to validate the Microbot power subsystem concept and provide useful insight into the design consideration for a micro-fuel cell power system. The system was designed around three air-breathing micro
PEM fuel cell provided by Dr. Freidrich Prinz and Dr. Tibor Fabian from Stanford
Univeristy. Air-breathing fuel cells derive their oxidizer from the air instead of a pure
Chapter 2: Fuel Cell Power System 29
oxygen source. Because the earth's atmosphere is only about 20% oxygen, there is a decrease in power density and energy generation efficiency due to the lower oxidizer concetration [38]. Research has shown that the use of pure oxygen instead of air increases PEM fuel cell power generation by 30% [42].
Therefore, the fuel cell efficiency values calculated from these experiments should be considered conservative.
The fuel cells are fed pure, humidified hydrogen from a pressure regulated compressed gas tank at 5 psi. The fuel cells are arranged in a series stack configuration to increase the output voltage. The hydrogen input and output tubes are in a parallel configuration to ensure that each cell recieves hydrogen at the same pressure. Figure 8 shows an image of the fuel cells and the humidification device. The fuel cells produce a varying voltage and current output depending on the internal state and external conditions of the cells. To maintain a constant voltage during operation, the system uses a low voltage regulator (LVR) to condition the fuel cell output.
Fuel Cells
Humidifyin
Beaker
1
AHy beaker.
Figure 9 shows an operational schematic of the power system prototype.
The electricity outputed by the first LVR, refered to in Figure 9 as LVR1, is sent down two
30
Chapter 2: Fuel Cell Power System
paths: to the PIC microcontroller and the MOSFET gate, and to the lithium-ion battery charger. The function of the PIC and MOSFET is to regulate the actuation of the DEA
by opening and closing the transistor gate. The gate allows the Li-ion battery to power the DEA prototype. The fuel cells charge the battery via the charging circuit. The connection between the LVR1 and the charger is broken in the diagram to indicate that the charging circuit only begins the charging process after the battery voltage has dropped below a specific threshold.
The MOSFET separates the low and high voltage sections of the power system prototype. The DEA requires voltages up to 10 kilovolts to operate, so a miniture high voltage converter (HVC) is used to generate the proper voltage. However, the HVC is unregulated and the DEA requires a precise voltage to actuate without any risk damage.
Therefore, a second low voltage regulator (LVR2) is selected to control the exact voltage that enters the HVC. The Li-ion battery, with assistance from the fuel cells via the battery charging circuit, powers the DEA. A battery is used because the fuel cells can not provide the peak power required by the DEA.
H
2
Fu Cell LVR1
By +dcs
Water + Heat
Charger PIC
Li Bat MOSFET
LVR2 HVC
Gate
Figure 9: A schematic of the fuel cell power system prototype.
Work
Output
DEA
31
31
Figure 10 shows the experimental prototype. Note that a larger hydrogen tank was used in the experiments and the water hydration device is not shown. Figure 10 illustrates the relative size of the power system components. The fuel cells, for example, could be further miniaturized and arranged into a stack to make the system more compact. Also, the low voltage electronics (LVR1, LVR2, PIC, and MOSFET) could be intergrated into a single circuit boards instead of individual boards. Finally, observe the small size of the HVC and the Li-ion battery. The current state of these technologies suggest that miniaturization of the entire system will be possible in the near future.
Appendix A contains a more complete description of the design and component specifications of the fuel cell system prototype.
Figure 10: fuel cell power system.
The prototype device in Figure 10 operates in the following manner the fuel cells are supplied hydrogen at a constant pressure of approximately 5 psi. The fuel cells generate electricity that powers the PIC, MOSFET, and charges the battery The PIC is programmed to open the MOSFET gate for 1.25 seconds and then close it for 10 seconds.
While the gate is open, the lithium-ion battery and the fuel cells power the HVC at the voltage determined by the LVR2. The high voltage signal produced by the HIVC is then
Chapter 2: Fuel Cell Power System
32
used to actuate the Microbot DEA and produce mechanical work. Measurements are taken during operation to evaluate the system performance.
The following section presents the power system experiment results and discusses the system performance.
2.2.2 Experimental Results and Analysis
The goals of the fuel cell prototype experiment are to demonstrate that a hydrogen fuel cell power system can continuously power the Microbot concept and to determine the performance of the technology. The experimental prototype is compared to the
Microbot power system requirements to demonstrate the feasibility of the hydrogen fuel cell power system concept.
The power system was tested in the following way:
* The fuel cells were activated by purging the anode flow channels with hydrogen.
" When an initial voltage increase was measured across the cells, the output end of the anode flow channels was blocked, allowing for dead-end operation mode. This operation mode is where the hydrogen passively diffuses into the fuel cell membrane.
" To increase the performance of the fuel cells, each cell was initially shorted to generate the maximum current and consequently increase the internal temperature.
* The cells were properly conditioned when the in-series open-circuit voltage of the stack was approximately 2.9 volts. They were then connected to the low voltage electronics.
* The fuel cell power system actuated the DEA every 11.25 seconds via a square wave determined by the PIC microcontroller and the MOSFET.
Chapter 2: Fuel Cell Power System 33
During the experiment the following measurements were taken:
1. The fuel cell voltage and current
2. The HVC output voltage and current
3. The Li-Ion battery voltage
4. The hydrogen consumption rate
Results
The experimental system demonstrated continuous system operation for over 90 minutes. The experiment could have continued for a longer period of time, however it was ended because a sufficient amount of measurements were taken.
The fuel cell system operates in two different modes: when the battery charging circuit is off and when the charging circuit is on. The fuel cells experience different loads in these two cases. The two states are discrete and binary because the charger turns on when the battery voltage has fallen below a certain threshold.
The fuel cell system hydrogen consumption and power generation were measured for both the charging and not charging cases. Note that the hydrogen flow rate into the cells can be considered a measurement of input power because the energy content of hydrogen, know as the lower heating value (LHV), is approximately 120,000
J/g [301.
Table 3 summarizes the experimentally measured power values. The hydrogen flow rate was measured by recording the size and rate of hydrogen bubbles flowing through the humidifier. See Appendix A for a more complete explanation of the hydrogen measurement.
Chapter 2: Fuel Cell Power System
34
Table 3: The experimentally measured power values for the system.
Measurement
Hydrogen Flow Rate, Not Charging
Hydrogen Flow Rate, Charging__
Fuel Cell Average Output, Not Charging
Fuel Cell Average Output, Charging
DEA Peak Power Input
DEA Average Power Input
Power
3.8 micrograms/sec
6.4 micrograms/sec
280 mW
560 mW
320 mW
15 mW
The fuel cells still power the low voltage electronics when the battery charging circuit is off and the DEA is not actuating. When the DEA actuates, the battery's voltage drops significantly, causing the battery charger to turn on. When this happens, the fuel cell sees an increased load for the 1.25 second actuation period. Figure 11 contains plots of the fuel cell stack power output for the two charging conditions. Notice that the square pulses in Figure 11(a) are 11.25 seconds apart. Also note that in Figure 11(b), the charging case, the DEA actuation has no affect on the fuel cells' output because the battery charging circuit is constantly on during this period.
FC Power Output: Not Charging
1 1
Charging
0.8 0.8
.) d-0.4
0.2
....
...
0.6
cL 0
4
............
0 0.2 .....
0 10 time (s)
(a)
20 0 10 time (s)
(b)
20
Figure 11: The fuel cell power output (a) without charging the Li-ion battery and
(b) with charging.
Chapter 2: Fuel Cell Power System
35
Figure 12 presents a plot of the DEA input power signal. The power input signal to the DEA is composed of an approximately 320 mW spike and
1 st order system dynamics. This signal is the product of the HVC and DEA dynamics. The negative portion of the signal can be ignored because it is the result of the capacitive elements in the DEA discharging after the actuation [39]. Excluding this negative power component, the average power into the DEA is approximately 15 mW (see Table 3).
DEA Power Input
0.35
0 .3 --
---- ------------
-0.15
0.1
0 5 10 time (s)
15
Figure 12: The DEA power input during two actuations.
20
To examine the system performance, component efficiencies were determined.
The LVR efficiency is current-dependant and is published by the manufacturer [31]. The
HVC efficiency was measured to be approximately 25%. The fuel cell efficiency is determined by calculating the percent of the chemical energy that is converted into electrical energy:
7F =
FC CH e x
FC
(120,000 Cg)
Chapter 2: Fuel Cell Power System
(2.2)
36
where 7 lFC is the fuel cell efficiency, PFC is the power produced by the fuel cell, rhH, is the mass flow rate of the hydrogen fuel in grams per second, and 120,000 J/g is the lower heating value (LHV) of hydrogen [2]. The efficiency value is higher when the battery is charging. This is because as more power and current is produced the internal temperature of the fuel cell increases, boosting the cell efficiency by decreasing the internal losses.
The DEA efficiency is calculated by the ratio of mechanical energy out and electrical energy in to the DEA device: in cycle
X100%
(2.3) where qDEA is the DEA efficiency, mDEA is the DEA mass, 6 grams in this case, and sDEA is the specific energy produced by the DEA during each actuation as a function of DEA mass, a value of 0.002 J/g for this prototype [39]. J, is the input power stated in Table 3 and tcycle is the period of each actuation cycle, 11.25 seconds. A DEA efficiency of 5.4% was calculated for this experiment. This value is comparable to other published efficiency values for these actuators [39]. Table 4 summarizes the efficiency values.
Table 4: The power system efficiencies.
LVR, Low oad_(
LVR, High Load (>100mA)
HVC
Fuel Cell, Without Charging
Fuel Cell, While Chargig
DEA
Component Efficiency
50%
90%
25%
60%
74%
5.4%
Performance Comparison
The power system prototype results demonstrate that hydrogen fuel cells are a feasible power source for the Microbot concept. The required performance of a Microbot
Chapter 2: Fuel Cell Power System
37
fuel cell power system is predicted for reference mission to Mars in section 4.1 and summarized below in Table 5. The reference mission requires the Microbot to execute
2000 hops over the course of a 7 earth days. The predicted performance values are compared to the averaged experimental performance results in Table 5.
Table 5: The averaged fuel cell prototype performance values and the predicted
Microbot power system performance specifications from section 4.1.
Exp. Prototype (Avg.)
Hop Period
Fuel per Actuation
Average DEA Work
11.25 sec
Fuel Cell Efficiency 64 %
Energy Generation Rate_ 0.42 W
Fuel Consumption Rate 5.1 micrograms/sec
58 micrograms/actuation
0.01 J/actuation
Microbot Concept
259 sec
60%
0.13 W
1.8 micrograms/sec
58 micrograms/acuation
0.46 J/actuation
The period of time between each actuation is 20 times greater for the reference mission case. This fact explains why the average fuel consumption rate is lower for the
Microbot concept. Between hops the Microbot consumes some fuel taking sensor measurements and transmitting data, however the mobility system requires a significant amount of power when actuating. The experimental prototype actuates more regularly and thus consumes more energy on average.
The experimental prototype DEA is not the proper scale for the Microbot concept: it can not produce enough energy to make a 100 gram Microbot hop. The prototype DEA device has to complete almost 46 actuation cycles to generate as much mechanical work as the DEA specified for the Microbot concept. The fuel consumption per actuation are approximately equal, however the experimental value also includes the small fraction of the power consumed by the electronics and does not produce as much mechanical work as the Microbot concept DEA. The prototype DEA has a considerably lower specific energy and efficiency than the Microbot concept actuator described in section 4.1.2.
Chapter 2: Fuel Cell Power System 38
The experimental fuel cells performed as well as the Microbot concept fuel cells.
Efficiencies as high as 74% were measured for the experimental fuel cells, exceeding the
60% efficiency selected for the Microbot concept power system (see section 4.1.2). The average power output of the fuel cells in the experiment is over 3 times greater than the power output predicted for the Microbot concept. The fuel cells designed for the
Microbot should be sized to the average power demands of the robot and a trickle charged batteries should be used to provide peak power. The cells should also be compact and very lightweight. This can be achieved if the cell structural components are composed of high-strength carbon fiber and plastics.
Discussion
The experiment demonstrates that the currently available fuel cell technology can perform sufficiently for the Microbot system. Note that the fuel cells in the experiment were air-breathing. If pure oxygen was used instead, the power generation performance would improvement by 30% [42].
The power regulation electronics efficiency values presented in Table 4 are promising, however improvements still needed to be made. For example, the current micro HVC technology is optimized for volume and weight, not for performance. A minimum efficiency requirement of 60% is calculated for the HVC (see section 4.1.2).
The power system prototype is currently too large and heavy for the Microbot system (see Figure 10). Integrating the different components through careful design can greatly decrease the system mass and volume. Also, arranging the electronics onto a single circuit board reduces the system complexity and increasing the efficiency. A goal mass for the power system, excluding the fuel, is less than 40 grams (See section 4.1.3).
The hydrogen storage method was not directly addressed in this work. A large compressed gas tank containing hundreds of liters of hydrogen was used. As discussed in
Chapter 2: Fuel Cell Power System 39
section 2.1.2, there are a number of different options for hydrogen fuel storage, each with their advantages and disadvantages. The most promising technology currently for miniaturization is the metal hydride technology because it does not require high pressures or additional processing components [52]. In the future, new hydrogen absorbing technologies such as carbon nanotubes may allow for further miniaturized and increased performance [11]. As indicated in section 4.1.2, a minimum hydrogen gravitation storage efficiency of 3.5% is specified for the Microbot system.
The power system performance can be optimized by closely monitoring the internal states of the fuel cells. Sensors should be used to measure the fuel pressure and flow rate, the internal temperature of the cells, and the humidity of the electrolyte. With the use of MEMS sensors, a miniature fuel cell monitoring system could be integrated into the Microbot power system design.
Finally, management of the fuel cell system byproducts must be addressed. The power system produces two byproducts: water and heat. The water byproduct must not be released to prevent contaminating the sensitive scientific measurements. This is especially a concern on the subterranean Mars mission where one of the main objectives is the search for signs of water deposits. Therefore the water byproduct must be collected and stored onboard the robot or recycled and used to humidify the fuel cells.
The heat byproduct can be used to maintain the Microbot's operating temperature.
Due to its lack of atmosphere, Mars experiences daily temperature swings of -120 'C to
20 'C [24]. The power system heat byproduct could be used to warm the Microbot.
Ideally, the hopping rate of the Microbot can be synchronized with the rate of heat loss to ensure metabolic control over the Microbot's temperature. The Microbot thermal model has been analyzed in detail [6].
Refer to section 4.1 for an analysis of the Microbot power and mobility system designs and a further discussion of the power system performance values.
Chapter 2: Fuel Cell Power System
40
CHAPTER
3
The objective of this chapter is to analyze the mechanics of the Microbot-terrain interactions. First, the amount the Microbot sinks into a deformable terrain during hopping and bouncing is examined through a terramechanics analysis. Next, the hopping energy lost in the terrain due to soil deformation is calculated. Finally, experimental results are presented to verify the theoretical analyses.
The analyses indicate that that the Microbot will sink into the terrain less than
15% of its diameter and that approximately 3% of the hopping energy is dissipated in the terrain. These values are experimentally verified and the results are within 40% of the predicted values.
In this section, the performance of the Microbot in deformable soil is modeled with terramechanics equations. The first analysis calculates how far the Microbot will sink into a Martian soil due to gravity or a collision after a hop. This value helps to determine if the Microbot will become trapped or unable to roll over the deformable
Martian terrain. The second analysis calculates the amount of hopping energy dissipated in the terrain through soil deformation. This analysis is used to estimate an efficiency value for the power and mobility system level study in section 4.1.
For these analyses, the terramechanics equations derived for ground vehicles by
Bekker are adapted for a spherical geometry and applied to the Microbot system [20].
Chapter 3: Microbot-Terrain Interactions 41
3.1.1 Terramechanics
The terrain interaction analysis presented here focuses on the static and dynamic interactions of the Microbot with a soil. These interactions are a fundamental design consideration for the Microbot locomotion system [20, 3].
Terramechanics is the study of soil properties, specifically how vehicles interact with different types of terrains [3, 20]. In this analysis, terramechanics is first used to determine to what depth a stationary Microbot will sink into a soil. This value is a function of the Microbot's geometry and mass and the soil parameters. This depth is solved for by iteratively calculating the vertical component of the force exerted on the robot by the terrain, N, for different sinkage depths until the terrain force is equal to the weight of the Microbot, mg. This analysis ignores the affects of friction on the
Microbot. Friction is not considered because the Microbot shell is assumed to engage the soil and thus soil compaction is the dominant force generating mechanism [26,57].
The value of N is the integral of the stress distribution in the soil over the surface area of the Microbot in contact with the soil:
N = J-(O) cosO dA surface
(3.1) where the normal stress in the vertical direction, o(0) cos 0 , is a function of the angle from the axis perpendicular to the terrain, 0 [26]. As shown in Figure 13, r is the radius of the Microbot, z is the sinkage depth, and 0, is the angle that defines where the
Microbot makes contact with the terrain.
Chapter 3: Microbot-Terrain Interactions 42
S01
Figure
A diagram of the Microbot-terrain
analysis.
The equation for stress in the soil in this analysis is: o-(0) =(k + k
2
Docal)(
DJ (cos0 -cos0
(3.2) where k
1 and k
2 are soil properties, Diocai is the diameter of the circular contact area between the ball and the terrain, and n is a soil-dependant sinkage coefficient constant
[26]. This stress is substituted into Eq. (3.1) to derive the normal force exerted on the
Microbot as a function of sinkage:
N =- k k2ic1 Dioca
0 0 osO-cos
1
)" cos.[-r 2 sin Gd6 d# ~ (3.3)
43
Chapter 3: Microbot-Terrain Interactions
# is the longitudinal direction on the where the angle 01 defines the sinkage and area of the
by iteratively adjusting the sinkage the stress distribution over the surface
Microbot sphere. Eq. (3.3) integrates
Microbot sphere in contact with the soil. The static sinkage is determined depth until the normal force (N) with Eq. (3.3) equals the gravity force (mg).
A similar sinkage analysis can be made to determine the sinkage depth from a dynamic collision between the Microbot and a deformable resulting terrain. When the
Microbot lands after an initial hop, its kinetic energy is converted into soil deformation, heat, and kinetic energy in the form of a bounce. However, in the highly plastically the fraction of the total energy converted into deformable flour-like soils of Mars, bouncing can be ignored [5, 47].
Preliminary laboratory experiments suggest that very limited or no bouncing will occur after the Microbot collides with soft soil. Hence for his analysis, the amount of energy spent deforming the soil, Edeform, is approximately equal to the Microbot kinetic energy at impact. This assumption, as well as the fact that friction is ignored, results in an upper bound value for the sinkage.
Figure 14 shows an image of the
Mars rover Opportunity's wheel stuck in the soft Martian sand.
Figure 14: A Opportunity rover wheel stuck in a soft Martian sand pile [37].
Chapter 3: Microbot-Terrain Interactions
44
Edeform is calculated with the integral of the normal force exerted by the soil on the
Microbot over the sinkage depth into the soil:
Edefo,,=
ZD
JN(z) dz
0
(3.4) where
Edeform is the soil deformation energy, zD is the ultimate sinkage depth of the microbot, and N(z) is the normal force exerted by the terrain as a function of sinkage.
For a known
Edeform
, a sinkage depth can be found by solving Eq. (3.4) iteratively.
As discussed above,
Edf,,r is approximated as the Microbot kinetic energy at impact,
Ecoision .
The kinetic energy at impact is approximately equal to the maximum potential energy during a 1 meter high hop.
Eolliion = mgh (3.5) where m is the Microbot mass, g is the gravity on Mars, and h is the 1 meter hop height.
Table 6 summarizes the equation parameters used to calculate Microbot sinkages. The soil properties used in this analysis are for dry sand with very low cohesion [26,57].
Table 6: The terramechanics analysis parameters used in this analysis [26].
Parameter
Microbot Radius
Microbot Mass
Hop Height
Gravitational Acc.
Soil Property
Soil Property
Sinkage Coefficient
_
Variable r m h k n
Value
50 mm
100 grams
1 m
3.69 m/s 2
0.9 kPa
1523 kPa/m
1
Chapter 3: Microbot-Terrain Interactions
45
Using these parameters and Eq. (3.1) (3.5), the sinkages presented in Table 7 are calculated. These values are upper bound sinkages because the fraction of the energy converted into bouncing and heat through friction is ignored.
Table 7: Microbot sinkage values predicted by the terramechanics analysis.
Sinkage Case
Static
One Meter Fall Height z
1.2mmn
17.1 mmin-
0
12.60
48.80
3.1.2 Hopping Efficiency and Terrain Energy Dissipation
The objective of this section is to calculate the amount of energy lost in the terrain due to hopping. For a known robot mass, volume, hop height, and terrain type, the energy dissipated through soil deformation during hopping can be determined. Using the terramechanics equations presented in the previous section, an energy dissipation of
0.01 J is calculated for a Microbot performing 1 meter high hops on Mars.
The energy lost during hopping is used to calculate the hopping efficiency.
Hopping efficiency is defined as the percent of the energy stored in the bistable device that is converted into kinetic energy during the hopping process [19]. This value is an important consideration for determining the size and energy requirements of the hopping mobility mechanism.
Energy losses during hopping are due to a number of factors [19]:
1. Energy dissipation in the bistable mechanism
2. Energy lost due to the mechanical linkages in the hopping mechanism
3. Slippage and friction between the hopping foot and the terrain
4. Deformation of the soil due to the hopping action
Chapter 3: Microbot-Terrain Interactions 46
For this analysis, the efficiencies due to the bistable device and hopping mechanism, losses (1) and (2), are decoupled and considered separately. This simplification can be made because these two losses are a function of the mobility mechanism design and not the robot-terrain interaction. Section 4.1.2 discusses losses due to the mobility mechanism.
This Microbot hopping efficiency is therefore considered a function of the energy dissipation between the Microbot and the terrain, Edissipate , losses (3) and (4). The energy dissipated can be estimated with a similar method to the dynamic collision analysis presented in section 3.1.1. This analysis approximates the energy as the normal force exerted by the terrain, N(z), integrated over the sinkage distance spanning the static deformation sinkage before the hop, z,, to the sinkage due to the thrust force of the hopping foot against the ground, z,.
Edissipate =
ZT
JN(z) dz
Zs
(3.6)
For this analysis zs is the static sinkage value calculated in section 3.1.1, 1.2 mm.
The thrust force sinkage is calculated with an iterative method similar to the one presented in section 3.1.1. To find z,, the soil normal force in Eq. (3.3) is iteratively calculated for different sinkage values until it is equal to the force exerted by the
Microbot on the ground during the hopping action. This force is equal to the force of gravity plus the thrust reaction force from the hopping action, mg + ma,h,,,. The thrust acceleration, athrust
, is approximated as the following: a
2gh
(3.7)
Chapter 3: Microbot-Terrain Interactions
47
where g is the gravity on Mars, h is the 1 meter hop height, and at is the period of time that the hopping foot is in contact with the ground during an actuation. This period is estimated as 0.05 seconds from prototype hopping mechanisms [27].
Using the equations parameters in Table 6, a thrust sinkage of 5 mm is calculated.
By applying this value to Eq. (3.6), the hopping energy dissipated in the soil for a 1 meter hop height, Edissipate, is calculated as 0.01 Joules or approximately 3% of the total hopping energy. Hence, this method predicts a hopping efficiency for the Microbot concept on
Mars of 97% due to soil deformation. 97% of the energy produced by the hopping mechanism is converted into kinetic energy. Fiorini and Burdick measured a hopping efficiency of 70%. However, their analysis is for an undisclosed hop height and terrain type. Their analysis also includes the energy losses in their robot's hopping mechanism, a 6-bar linkage with linear springs, and friction [19].
Simple laboratory experiments were conducted to validate the terramechanics and energy dissipation calculations presented above. Although the Martian environment can not be exactly simulated on Earth, an approximation of the conditions is useful for verifying the theoretical analysis. The experiments presented here are a starting point for further investigation.
3.2.1 Experimental Setup
The Martian soil was simulated with two different materials: fine sand and packed wheat flour. While the soil found on Mars varies over the planet, these selections are reasonable approximations for common types of top soils [47]. The sand and flour were held in plastic buckets. The depth of the simulated soil in each container was at least 15 cm.
Chapter 3: Microbot-Terrain Interactions
48
To correct for the different gravitational accelerations on Earth and Mars, the
Microbot mass was adjusted for the laboratory experiments. The gravitational acceleration on Earth, gEarth is 9.8 M/s
2 and the gravitational acceleration on Mars, gmars, is 3.69 m/s
2
, approximately 40% of the value on Earth. To correct for this difference, a
40 gram Microbot mockup was selected to represent the 100 gram Microbot concept.
Ignoring air resistance, the gravitational force and potential energy of the Microbot concept on Mars and the Microbot mockup on Earth are the same:
Mmockup g
Earth =
M concept g Mars
(3.8)
Mmockup g
Earthh = mconcept g
Marsh
(3.9) where mmocku, is the mass of the mockup, 40 g, mconcept is the mass on the Microbot concept, 100 g, and h is the vertical position of the Microbot relative to the terrain surface in meters. The kinetic energy of the Microbot at impact is approximated as the maximum potential energy of the Microbot during a hop. Therefore, Eq. (3.9) illustrates that the energy at impact of the Microbot concept and mockup are approximately equal.
The Microbot mockups are hollow polypropylene balls purchased from
McMaster-Carr (part #: 3748K18). Although the mechanical properties of the Microbot concept's exterior have not yet been investigated, it is assumed that the Microbot's outer shell will have a high stiffness and low damping to allow for bouncing and rolling after each hop. The stiff polypropylene shells are therefore a suitable candidate for these experiments.
The static sinkage experiments were carried out by lightly placing the sphere on a level sample of sand and flour. For the collision tests, the Microbot mockup was dropped from a 1 meter height onto the two different materials.
Chapter 3: Microbot-Terrain Interactions
49
For the hopping efficiency experiment, the force of a Microbot hop on sand was approximated. To achieve this, the hopping mechanism from a Microbot prototype was configure so that it could be manually actuated [27,28] The amount of force produced by the hopping actuator is determined by the deflection of the power springs. To simulate the hopping force, the mobility system was actuated against the sand with approximately the same spring deflection as a normal hop.
To determine the sinkage depth for each of these experiments, the diameter of the resulting impact crater was measured. The diameter was measured instead of the sinkage depth to reduce estimation errors. For each of these experiments, a number of trials were run and the results were averaged.
The sinkage depth, z, can be calculated from the indentation diameter in the following way: z=r(1-cos(arcsin(f))) (3.10) where r is the Microbot mockup radius of 50 mm and D is the diameter of the circular indentation in the soil.
3.2.2 Experimental Results
Table 8 summarizes the averaged results of the experiments described above. The sinkage for each experiment, z, was calculated with Eq. (3.10). The terrain contact angle defined in Figure 13, 01, and the percent error relative to the predicted sinkage values were also calculated. The results from the sand experiments are presented because the theoretical analysis used sand soil parameters.
Chapter 3: Microbot-Terrain Interactions 50
Table 8: The results of the Microbot-terrain interaction experiments in sand.
Experimental
I
Sinkage Case
Predicted
0z z
Static
1.2 mm 12.6'
One Meter Drop jight 17.1 mm 48.80
Thrust Force Sinkage
5.0 mm 25.8
1.6 mm j 1
1450
12.0 mm 40.50
4.2 mm 23.6
% error
_ _
+25 %
_-42_%
-19%
Figure 15(a) shows a top view of the sinkage indent in flour and an outline for clarity. Figure 15(b) shows an approximate orthographic view of the indent.
Figure 15: Two views of the impact indents made by the Microbot mockup in flour: (a) Top view of the indent with a ruler and indent outline
(b) an orthographic view of the indent.
The values in Table 8 have a non-negligible percent error in comparison with the predicted values. The two experimental results that involve dynamic interactions, the collision and hopping force cases, are less than the predicted values because the theoretical analysis does not consider the energy dissipated due to friction. Another reason for the discrepancy is that the soil properties used in the analysis may not match the properties of the sand used in the experiments
The predicted and the experimental sinkages are less than 15% of the Microbot diameter. This level of sinkage suggests that the Microbot will not become fully entrapped in the Martian sand. Depending on the weight distribution and design of the
51
Chapter 3: Microbot-Terrain Interactions
hopping foot, self-righting may be possible. The results also suggest that due to the rolling resistance caused by the soil deformation, the Microbot will not always be able to roll freely after each hop. However, this prediction depends on the slope and composition of the terrain.
Chapter 3: Microbot-Terrain Interactions 52
CHAPTER
4
The objective of this chapter is to determine if the Microbot concept can successfully complete a reference mission on Mars. First, a power and mobility system model is created with conservative efficiency values and evaluated for the reference mission. This analysis connects the hopping mobility and fuel cell power system research presented in this thesis.
The second section of this chapter focuses on the Microbot mobility system.
Dynamic simulation software is used to examine the performance of the hopping mobility method on a Martian terrain.
The results of the analyses demonstrate the power and mobility feasibility of the
Microbot concept. A Microbot with the design specifications summarized in Table I is shown to able to complete a reference mission of over 2000 hops on Mars. The mobility simulations indicate that the Microbot hopping, bouncing, and rolling mobility system will have some success traversing lava tubes and overcoming obstacles.
4.1 System Level Feasibility
The feasibility analysis in this section focuses on the integrated mobility and power systems of the Microbots. Other issues, such as the sensors, team navigation and thermal regulation present important technical challenges [12,13,6]. However, these areas are beyond the scope of this analysis.
Chapter 4: Feasibility Analysis 53
To evaluate the feasibility of the Microbot system concept, a sample Martian lava tube reference mission is considered. In this mission, the Microbots would be deployed on the surface of Mars, travel a kilometer over rough terrain, and penetrate 500 meters into a cave while collecting scientific data (see Figure 16)
[27]. From this reference mission profile, it is predicted that the Microbot must perform as many as 2000 hops on
Mars to successfully complete the mission. This number is selected as an upper bound value to account for the Microbot having to circumvent obstacles and steep inclines.
The mission duration is selected based on the desired hopping rate and the amount of time required to take sensor readings. For this study, the reference mission objectives are achieved in 7 Earth days, or about 6.8 sols (Martian days).
Figure 16: A concept drawing of Microbots entering a cave on Mars [13].
4.1.1 System Model
Two key questions are addressed in this analysis: 1) can the system generate enough power to successfully complete the mission and 2) can the Microbot carry enough fuel to travel the required distance without exceeding the size and weight design specifications? For this analysis, the desired Microbot mass is 100 grams and the diameter is 100 mm. These specifications are summarized in Table 1 in section
1.3.
54
Chapter 4: Feasibility Analysis
Figure 17 illustrates the energy conversion model used in this analysis.
QHO
Q Q
Gas
Storage inetic ne
Fuel J12
Cell
Power
Regulation
73 Mobility
System ave
P2
Electronics and Sensors
Figure 17. The Microbot energy conversion model.
In Figure 17, the symbols are:
1. 1, is the efficiency of the fuel storage system, the ratio of the mass of the fuel to the mass of the entire fuel storage device.
2. r72 is the efficiency of the fuel cells, expressed as the percent of the lower
3.
4. heating value (LHV) of hydrogen that is converted into electrical energy.
The LHV of hydrogen is approximately 120,000 joules per gram [2].
17
3 is the electrical efficiency of the power regulation subsystem.
r74 is the mobility system efficiency of converting electrical energy into
5.
6. kinetic energy.
Pve
Q is the average power draw of the electronics.
is the heat generated in the energy conversion processes. This heat is necessary for the thermal viability of the Microbot system in the cold
Martian environment [6].
The following section analyzes these model parameters in greater detail.
Chapter 4: Feasibility Analysis 55
4.1.2 System Efficiencies
1) Fuel Storage
The efficiency of the fuel storage system, qj, is equal to the percent of the fuel storage device weight that is fuel. This dimensionless number is a useful metric for comparing different types of fuel storage systems. Section 2.1.2 presents a comparison of hydrogen storage technologies. The PEM fuel cell fuel and oxidizer, hydrogen and oxygen, react in a 1:8 mass ratio.
Hydrogen is not as simple to store as oxygen due to its extremely low atomic weight. The current best methods for storing hydrogen are in chemical or metal hydrides or as liquid hydrogen [52]. Since using liquid hydrogen and chemical hydrides add substantial system complexity, a metal hydride was chosen as the storage media.
Currently one of the best metal hydride technologies is sodium alanate, which stores hydrogen with a weight efficiency of 5.5% [52]. Because additional containment and pressure regulation are required, this analysis uses a conservative estimate for the overall hydrogen storage percent weight of 3.5%.
Oxygen has a considerably larger atomic weight and can be stored as a compressed gas. The oxygen can be stored at high pressures using high-strength carbon fiber composites. Considering the weight of the container and pressure regulation, a storage efficiency of 35% is estimated.
2) Fuel Cells
PEM fuel cells were selected because of their low operating temperature of 0-
120
0
C and efficiency as high as 70% [38]. Given the volume and mass constraints of the system, this analysis assumes a conservative energy efficiency (0
2
) of 60% for the fuel cell system. Figure 18 shows an example of a miniature PEM fuel cell.
Chapter 4: Feasibility Analysis 56
Figure 18: A miniature PEMfuel cell [1].
3) Power Regulation
The Microbot system requires 3-5 volts for its electronics and communications and 7-10 kilovolts for the DEA mobility elements. Using current technology, the low voltage regulator can have a conversion efficiency as high as 96% [31]. A miniature high voltage converter can have an efficiency as high as 85% at 10,000 volts (see Figure
19)
[41]. Because these regulators may not always be operating at their optimal point, more conservative efficiencies are used: 90% for the low voltage regulator and
60% for the high voltage converter.
Figure 19: A miniature high voltage DC-DC converter [15].
4) Mobility System
The efficiency of the mobility system is broken down into three main components: the DEA, the bistable mechanism, and the mobility action.
The main energy losses in the DEA are due to electrical current leakage through the elastomeric film and viscoelastic effects [40]. Current DEA efficiencies are on the order of 5-10%
57
Chapter 4: Feasibility Analysis
[39]. This value will be substantially increased as new elastomer materials are developed. For this analysis, it is assumed that a DEA efficiency of 20% is a reasonable performance goal for a 10-year development timeframe. See Figure 20 for a demonstration of the capabilities of current Microbot laboratory prototypes.
The bistable energy storage mechanism has an estimated efficiency of approximately 90%. The final component of the mobility system efficiency is the energy lost during the hopping action principally to soil deformation. Based on preliminary calculations in section 3.1.2, a thrust efficiency of conservative 90% is selected. The product of these three values, 77 4
, gives an overall mobility system efficiency of 16.2%.
Another performance parameter of the mobility system is the specific energy, eDEA
, the work output per actuation cycle as a function of the DEA mass. Assuming space-quality DEA fabrication and development, this value is estimated to be on the order of 0.1 J/g [39]. This number is used to determine the mass of the mobility system.
Figure 20: A current prototype of the Microbot system capable of hopping. It contains a DEA actuator, energy storage, and integrated electronics [27].
5) Electronics and Sensors
Using MEMS sensors, ultra-high efficiency components, and distributed computation, the power demand of the electronics will be very low [12,13]. For example, currently available commercial off the shelf wireless sensor platforms are able to receive,
58
Chapter 4: Feasibility Analysis
process, and transmit data over high frequency radio communication with less than 50 mW of power [13,14].
The largest power consumption in the electronics system is assumed to be the communication subsystem. Because neither the sensors nor the communication systems must be on at all times, a duty cycle can be implemented that minimizes power consumption. Assuming a communications power consumption of 50 mW operating with a light duty cycle and the use of MEMS sensors and high efficiency integrated electronics, an average power demand of 100 mW is estimated.
Table 9 summarizes the efficiency values discussed above.
Table 9: The Microbot subsystem efficiencies used in the system level study.
Subsystem
Hydrogen Storage
Oxygen Storage __
Fuel Cells
Low Voltage Regulation
High Voltage Regulation
DEA Actuator
Bistable Mechanism
Hopping Action
Efficiency
3.5%
35%
60%
90%
60%
20%
90%
90%
4.1.3 System Analysis Results
The goal of the power and mobility feasibility analysis is to determine if a 100- gram Microbot can successfully perform the reference mission given the efficiency values in Table 9 and system model in Figure 17. The mass of the Microbot consists of several components. The first component is the mass of the electronics and sensors, the
Microbot shell and internal structure, and the fuel cells. The masses of these items are essentially a constant as they are not a function of the mission range or duration. In this analysis a total value of 40 grams is used for these items. This number is based on current lab prototypes [27].
Chapter 4: Feasibility Analysis 59
The other major mass elements are the mobility mechanism, the fuel and fuel storage containers, and the actuator. The amount of fuel (hydrogen and oxygen) required is a function of the total energy required for the mission. The total energy, Eotal, required from the fuel cells is:
Ettal = Eeectonic+EKE
where
Eeeconic is the energy consumed by the electronics and communications systems and
EKE is the mobility energy required to perform the mission. The energy required for the electronics is:
Eeectonic = Pavetmission
(4.2) where Pave is the average power requirements of the electronics and t,iion is the length of the mission.
Assuming limited bouncing and rolling, the kinetic energy required to perform
Nhops hops is a function of the total system mass, the mobility system efficiency (7
4
), gravity, and the assumed hop height (one meter):
(4.3)
EKE
= q
4
Nh opmgh where m is the mass of the Microbot, g is gravitational acceleration, 3.69 m/s 2 on Mars, and h is the hop height. The energy required to orient the Microbot hopping foot is not considered here because it is less than 1% of the total hopping energy [27].
Referring to Figure 17, the fuel mass consumed to produce a given amount of electrical energy (Eoal ) is given by:
Chapter 4: Feasibility Analysis 60
MH total44)
2 q23LHVH2 where m., is the mass of the hydrogen in the fuel storage device and LHVH, is the lower heating value (LHV) for hydrogen at the Microbot operating temperature (120 kJ/g) [2].
The actuator mass is determined by the mobility system performance, the
Microbot mass, and the jump height requirement: mDEA -
17 mgh bistable
17 hop eDEA
(4.5) where mDEA is the mass of the DEA device, ,bistabl, is the efficiency of the bistable device, the mechanical energy storage mechanism, r7hop is the hopping thrust efficiency, and eDEA is the specific energy output of the DEA per actuation. The mass of the Microbot
(m ) includes the mass of the DEA mobility system (mDEA ).
The number of possible hops, NhO,, can be calculated for a given kinetic energy and Microbot mass value:
N E KE hosmgh
(4.6)
Figure 21 shows the total number of possible hops as a function of total Microbot mass. These numbers assume a mission duration of 7 Earth days and a constant power consumption of 100 mW for the electronics and sensors.
Chapter 4: Feasibility Analysis 61
14000 - - -- ------------- q) 12000
16000
-- --
------- - ----------
--- - --- - ---
-------- I -------- - ------
100004 1- 1I-
0
E
4000 ---------
100
120 140 160
Microbot Mass (g)
180
Figure 21: A plot of the total number of hops on Mars during a 7 Earth day mission as a function of the Microbot mass.
The results of this analysis indicate that the Microbot is able to complete the mission objectives within the design specifications. The analysis predicts that a 100 gram
Microbot can perform 2334 hops on Mars. The results of the above calculations for a
100 gram Microbot are summarized in Table 9.
The electronics consume a significant fraction of the energy produced by the fuel cells during the course of the 7 Earth day mission. This fraction decreases as the total
Microbot's mass and the fraction of the energy consumed by the mobility system increases. Figure 22 illustrates this result.
Chapter 4: Feasibility Analysis 62
0
70 ----
100 20 140
Microbot Mass (g)
160 180
Figure 22: The percent of the fuel that is consumed by the mobility system during a 7 Earth day mission to Mars.
The analysis also calculates that the heat generated by the Microbot system,
Q, is 0.21 watts. This is a sufficiently high value to ensure that the Microbot maintains an acceptable temperature in the cold Martian environment [6].
Table 10: The Microbot design values determined by the system level study.
Design Parameter
Total Oxygen Storage Mass
Total Hydrogen Storage Mass
DEA Actuator Mass
Fixed Mass Items
Total Mass
Number of Hops
Mission Length
Q
10.21
Value
24.5g
3 0.5_g
5_g
40.g
1001
2334
7 Eatds
W
The Microbot power system in this study is compared to an experimental fuel cell power system prototype in section 2.2.2. This comparison validates the selection of fuel cells as the Microbot power sources and demonstrates the feasibility of the technology for this application.
Chapter 4: Feasibility Analysis
63
In addition to demonstrating feasibility through a system level analysis, feasibility can also be examined through simulations of reference missions. To be successful, the
Microbot exploration platform must be able to reach its goal destinations. Analyses of the mobility system can be conducted with a dynamic simulator tool. The goal of the mobility simulations is to examine the operation of the Microbot in realistic terrain and the affect of the mobility design parameters on performance.
The simulations were conducted with MSC Software's ADAMS dynamic simulation software. ADAMS allows the definition of mass properties, body forces, body constraints and interaction forces. A Mathworks Simulink model communicating with
ADAMS controls the hopping directionality. Refer to Appendix B for technical details of the simulations.
The Microbot interacts with the environment through hopping, bouncing and rolling on the terrain. Hopping is modeled as an impulse force between the Microbot and the terrain. The hopping direction depends on the angle that the impulse is applied relative to the Microbot's body.
Section 4.2.1 discusses the simulation objectives, how the terrain is constructed, and what assumptions and approximations are made. Section 4.2.2 presents a study of how the design parameters affect performance. The ability of the Microbots to overcome a rock pile obstacle is also examined. Finally, section 4.2.3 presents the skylight reference mission simulation.
4.2.1 Simulation Development
The objective of the following simulations is to examine how the Microbot mobility system performs in mission-specific locations. Although the Microbot
Chapter 4: Feasibility Analysis 64
specifications have been stated in previous work, these values are a product of preliminary analyses [13,27]. The simulations further examine the design specifications through mobility performance analyses. The simulations are also be a flexible tool for future Microbot research.
The terrain used in these simulations is assembled from a number of sources and constructed with several different software tools. The major features of the terrain are a hill, a rough plain, a lava tube with a series of skylight openings, and rock pile obstructions in the lava tube. Figure 23 shows the terrain.
160 meters
Lava Tube Opemi
Figure 23: A diagram of the simulation terrain.
The hill is generated by inverting a stereo camera image of the Martian crater
Endurance recorded by the Mars Exploration Rover (MER) Opportunity [33].
The surface roughness is derived from rock size distribution data from a Viking Landing site
[23]. The lava tube and rock piles are designed to closely resemble their counterparts on earth as shown in Figure 24. Very limited information exists about the conditions inside
Martian lava tubes, however orbiter images have recorded evidence of their existence in a number of areas on the planet [47]. Appendix C presents a more detailed technical explanation of the terrain constructed.
Chapter 4: Feasibility Analysis
65
Figure 24: A lava tube skylight in Idaho, USA [22].
The rock pile obstructions are located inside the lava tube cave directly underneath the skylights. This arrangement is selected to simulate the rubble from the cave ceiling collapse. The terrain's dimensions are approximately 280x160 meters in the x-z plane and the vertical elevation varies from approximately 5 meters to 18 meters at the top of the hill (see Figure 23).
Figure 25 shows a close up image of the simulated lava tube cave and rock piles.
Figure 25: The lava tube cave and skylights in the simulated
Mars terrain.
Chapter 4: Feasibility Analysis
66
Several approximations and assumptions are made in the design of the simulations. First, the force generated by the DEA mobility system is modeled as an impulse between the Microbot and the terrain. To simplify the control system, the
Microbot hops at 3 seconds intervals instead of sensing when the Microbot has come to rest. However, this period of time is normally sufficient for the Microbot velocity to drop to approximately zero. Appendix B presents additional details regarding the control system.
Due to limitations of the ADAMS software, a simplified contact model without plastic deformation is used to approximate the interactions of the Microbot with the terrain. The contact model generates forces between the Microbot and the terrain in the direction that opposes the relative motion of the two bodies. The resulting impact force is modeled as a nonlinear spring/damper system:
Fmpact =
-k(Ax)
2.2 b(Ai) (4.7) where k is the spring stiffness constant, b is the position-dependant damping coefficient, and Ax and &x are the relative displacement and velocity of the two bodies.
The force exponent of 2.2 is selected for materials that stiffen during deformation. Also, the ADAMS contact model is most robust with force exponents greater than 2.1 [34].
The friction force in the simulation is Columbic friction with a velocity dependant friction coefficient, p(Ai). The simulation's contact model parameters were estimated from laboratory experiments. The behavior of a Microbot mockup on compacted dry sand and rocks was observed (see section 3.2). The spring contact coefficient k is calculated from static plastic deformation test. The damping values, b, are selected so that the bounce heights in the simulation match the bounce heights measured in lab experiments. The friction coefficients, p , are selected so that the sliding and rolling
Chapter 4: Feasibility Analysis 67
dynamics after a hop match the results observed in the lab on the sand and flour soils.
Table 11 summarizes the contact model values.
Table 11: The impact contact model values.
Parameter k b
b (rock)
Lstatic
_10
p dynamic
Stiction Transition Velocity
Friction Transition Velocity
Value
240,000 N/m
N-s/n
0.5 N-s/m
2
0.15
0.01 M/s
0.1 m/s
The columbic friction model was used to simulate rolling friction. Laboratory experiments suggest that the Microbot will experience significant rolling resistance on the highly deformable soils of Mars. Rolling resistance is due to the deformable soil applying a moment to the rolling body [3]. However, ADAMS does not accommodate deformable bodies. Thus, rolling friction was not implement in the simulations. A solution to this challenge is to artificially increase the moment of inertia of the Microbot so that the rolling motion can be converted into sliding. The kinetic energy can then be dissipated in the Coulomb friction through the sliding motion. The moment of inertia was increased from approximately 0.00025 kg-m
2 to 1 kg-M
2
.
Although this is a significant approximation, simple simulations using the contact model values presented in Table 11 produce results that match well with experimental laboratory verification.
The rolling friction approximation was made to allow for the use of the ADAMS simulation tool. ADAMS is crucial for this research because it can simulate dynamics on large and complex terrains. A more appropriate method to simulate the rolling resistance would be to combine a finite element analysis (FEA) tool with a large-scale dynamic simulation tool. Each time the Microbot interacts with the terrain, the FEA software
Chapter 4: Feasibility Analysis 68
would calculate the position and forces on each particle of sand and determine the resulting plastic deformation of the terrain. However, this level of analysis is beyond the scope of this research.
The simulations presented in the following sections use the terrain, assumptions, and approximations discussed above.
4.2.2 Overcoming Obstacles
A simple simulation was constructed to examine the affect of design parameters on performance [28]. The simulation calls for the
Microbot to travel down a lava tube cave segment and successfully pass over a large rock pile in a limited number of hops.
This is a scenario that the Microbot will likely encounter on the Martian reference mission. Figure 26 depicts an illustration of this task.
Tunnel
Rock Pile
Figure 26: A Microbot traversing the simulated rock pile.
The simulated terrain was generated in Solidworks
CAD software as an assembly of individual solid bodies (see Figure 26). The rock pile is composed of 386 rocks of different sizes randomly grouped together into a pile approximately 5x4x0.85 m. Table
12 summarizes the rock size distribution in the pile.
The tunnel diameter is 5 m and its length is 60 m.
69
Chapter 4: Feasibility Analysis
Table 12: The rock size distribution in the rock pile obstacle.
Rock Diameter
10 cm
...
cm
40 cm
Quantity
62
222
102
Results and Discussion
A large number of simulations, over 150, were run. Microbot diameters of 5, 10, and 20 cm and hop heights of 50, 100, 150, and 200 cm were used. Each combination of hop height and diameter was simulated from a different starting positions spread over an approximately 2 meter area. The Microbot mass was fixed to 100 grams. In each simulation, the Microbot started approximately 2 m from the rock pile and had 14 hops to overcome the obstacle. This number of hops was selected because it allows the Microbot at least two tries to overcome the obstacle and it limits the length of the simulation to allow for a large number of iterations.
Figure 27 shows the Microbot success rate as a function of hop height. Success is defined as completely overcoming the rock pile. The trials that did not succeed had the following three outcomes:
1. Entrapment: A group of rocks traps the Microbot, rendering it unable to hop out.
2. Low hop height: All of the simulations with a 50 meter hop height were unable to hop over the rock pile.
3. Bouncing off: The Microbot hops in such a way that it bounces off the rock pile and lands a distance away and as a result can not complete the task in 14 hops. The consequence of bouncing away is not a failure per se but an undesirable delay.
Chapter 4: Feasibility Analysis 70
100.00%
90.00%
80.00%
70.00%-
60.00%
(50.00%---
40.00%
30.00%
10.00%
0.00%
-
-~E-
-
100
-
150
--
50
-
Hop Height (cm)
200
Figure 27: The success rates of all of the simulations as a function hop height.
Figure 28 illustrates the rate of entrapment for each combination of
Microbot diameter and hop height. Note that a greater rate of entrapment was found for the
Microbots with a diameter of 10 cm than a diameter of 5 cm. This is because although both size Microbots can become entrapped in a crevice underneath a rock, it is more likely that the 5 cm Microbot will be able to hop back out. The 10 cm Microbots are unable to extricate themselves because they continue to bounce of the entrapping rocks due to their larger size.
71
Chapter 4: Feasibility Analysis
100.00%
'$
80.00%
60.00%
40.00%
CL r 20.00%
-h
0.00%
Hop Height
Microbot
50 100 150 200
Dia. 5cm
50 100 150 200 50 100 150 200
10cm 20cm
Figure 28: The rate of entrapment as afunction of Microbot diameter and hop height
Figure 29 shows the rate of bouncing off of the rock pile for each combination of
Microbot diameter and hop height. Note that the majority of rates are approximately the same. This is because bouncing off of the rock pile has less dependence on the height of the hop or the size of the Microbot, but is primarily a function of where the Microbot contacts the rock pile and at what velocity. This occurrence does not necessarily cause a mission failure, but it does retard the progress of the Microbot through the cave.
100.00%
80.00%
60.00%
0
0 40.00%
0 20.00%
0.00%-
Hop Height
Microbot Dia.
50
100 150 200 50 100 150 200
5cm 10cm
50 100 150 200
20cm
Figure 29: The rate of bouncing off as a function of Microbot diameter and hop height.
72
Chapter 4: Feasibility Analysis
The results show that all trials with the lowest hop height, 50 cm, resulted in failure. This result suggests that a hopping robot can overcome a complex obstacle only if the hop height is greater than a characteristic height of the features on which it climbs, in this case approximately 0.85 m. This is because the Microbot can not settle on a rock ledge or face due to the lack of rolling resistance and high coefficient of restitution between the Microbot and the rock. It is possible that the Microbot can climb up smaller features with path planning and appropriately shaped ledges for it to stop on between hops.
In these simulations a hop height of 1 m leads to some success. Hence, hop height should be maximized. However, increased hop heights trade off with larger power consumptions and mechanism weights.
The results also indicate that a medium Microbot size results in greater entrapment. The rock pile was randomly assembled and is not an exact model of a real rock pile on Mars. However, it can be deduced that the maximum sized Microbot should be selected to minimize the chance of entrapment while still allowing for clearance in small cave openings.
The bouncing off cases are not a concern because they only retard the Microbot's progression. These cases could be improved or eliminated by more effective path planning.
4.2.3 Reference Mission Simulations
The objective of this section to present the reference mission simulation platform and the results of one study conducted with the simulation tool. The goal of the reference mission simulations is to demonstrate that the Microbots can successfully complete a
Martian lava tube exploration mission in a realistic simulation environment. The simulation consists of the large Martian terrain in Figure 23, a simple Microbot model,
Chapter 4: Feasibility Analysis 73
and a control system that allows for directional hopping. The control system also measures the Microbot's position and dynamics on the terrain environment. Appendix B summarizes the technical details of the simulation.
The reference mission simulation is also a research tool. The simulation is a stable platform for investigating and comparing different Microbot mobility system design concepts. Through repeated trials, the simulation has been shown to be robust to modifications. It does not crash when modification are made. This is a significant attribute because it allows simulation changes to be automated, increasing productivity and result output. Another benefit of the simulation is that it uses a terrain big enough for large numbers of Microbots to explore simultaneously. The simulation can also be used to investigate the success rate of the Microbots traveling long distances.
One example of a useful application of this simulation platform is the investigation of the optimal number of Microbot hopping directions. Although the directionality mechanism has yet to be designed, it is proposed that DEA actuators be used to tilt the Microbot [13, 27]. The optimal number of Microbot hopping directions is a significant unanswered research question.
The simulation platform was used to investigate this question. A simulation was created to evaluate the Microbot's ability to completing a 50 m traverse on a flat plain at a 370 angle from the z-axis shown in Figure 23. First, the paths were predicted for mobility system designs with 4, 8, 16, and 32 possible hopping directions. A prediction of the continuous hopping direction case, where the Microbot can choose any hopping angle to reach its goal, was also made. The path prediction criterion is to minimize the distance traveled by the Microbot given the limited number of possible hopping directions. The shortest path was selected for each number of possible hopping directions.
Chapter 4: Feasibility Analysis 74
The 50 m traverse was then simulated on the Mars terrain. The control system was modified to allow for 4, 8, 16, 32, or infinite (continuous) hopping directions. These directions were equally distributed among all 360' of rotation. The Microbot start position is in an open area adjacent to the lava tube. The goal position was located 50 m away at a 370 angle from the z-axis. The flat terrain between the start and end positions contains the roughness described in Appendix C. Figure 30 shows a comparison of the predicted and simulated trajectory results. Each simulation was run until the Microbot reached the goal position. Also, the Microbots came to rest between each hop due to the terrain interaction forces described above.
Predicted
45
40 END
Simulated with Rough Terrain
45
END
4dir. 35 -
E
30 -
25 d
2058
-- 16 dir
N15*odi
0
\
2.-32 d if
35 -
4 dir.
30 -
E
A
25
8 dir.>
1'
%\9
20 .16 dirr-X 32 dir.
0
15
" 4-oo dir.
10- 10-
5
0
-
0
START
20
X position (m)
40
5-
0F
0
STAR
20
X position (m)
40
Figure 30: The predicted and simulated path of Microbots with five different numbers of hop directions.
The results in Figure 30 show that the simulated trajectories match well with the predicted trajectories. The 4-direction mobility system, the system that can only go
Chapter 4: Feasibility Analysis 75
forward, back, left, and right, does not have enough directionality to efficiently reach the goal position. The 8-direction system is considerably better, and there is little difference between the 16, 32, and continuous direction systems.
Increasing the number of possible directions adds additional complexity, weight, and opportunity for malfunction and failure. Therefore, it is desirable to select the minimum number of directions that allow for adequate performance.
For this analysis, 8-directions provides sufficient directionality with a minimal number of hopping directions.
This analysis is one example of the possible uses for the reference mission simulation platform. The objective use is to simulate the complete Microbot Mars reference mission by evaluating the performance of hundreds of Microbots traversing the
Martian surface and then entering a lava tube through skylights. However, this simulation has not yet been done. Preliminary simulations have been conducted of single Microbots approaching and entering the skylights.
Figure 31 shows the trajectories of two preliminary Microbot simulations.
The maximum average distance traveled between hops in these simulations is 1.87 m, almost twice the distance assumed for the power and mobility system level study in section 4.1.
10meters
Figure 31: A diagram of two simulated Microbot trajectories over the surface, into the skylight, and down the lava tube cave:
(a) a traverse over the hill and (b) a traverse around the hill.
Chapter 4: Feasibility Analysis
76
After the simulated Microbots enter the lava tube via the skylights, they are instructed to progress down the length of the cave. During this process, they may become entrapped or experience the difficulties discussed in section
4.2.2. See Figure
32 for an image of Microbots in the simulated lava tube.
Although scientists do not know what is in an actual Martian lava tube, it is likely that the Microbots will encounter a number of obstacles. In the simulations, after the Microbots enter through the skylights, they must overcome a series of rock pile obstacles to reach the goal at the end of the cave.
Figure 32: Microbots hopping over rock piles in a lava tube simulation.
4.3 Design Tradeoffs
The Microbot system study shows that the system level power and mobility design is feasible. Using conservative efficiency values, it demonstrates that the
Microbots will be able to successfully complete a 2000 hop
Mars reference mission with the required design parameters. However, certain design tradeoffs should be considered.
The Microbot mass and power requirements are closely coupled.
If the sensors require more energy than predicted to operate, the Microbot will have to carry more fuel for a given mission duration. The greater quantity of fuel increases the system mass, thus requiring a larger hopping actuator. The coupled nature of the mobility system, power
77
Chapter 4: Feasibility Analysis
requirements, mission duration, and system mass should be considered when designing the Microbot.
Another area that is crucial to the feasibility of the concept is the Microbot mobility system design. The studies presented above suggest that the hopping mobility will have some success on the Martian terrain. The hopping direction study indicates that for this simulation, an 8-direction mobility system is an acceptable tradeoff between hopping mechanism complexity and path selection flexibility.
A hopping mobility system with more directions would require an increased number of components, resulting in an increased system mass. However, less hopping directions would not allow the
Microbot to choose a sufficiently direct route to its target position.
The performance of the Microbots will not only be a function of the mobility system, but will also be dependant on the condition of the terrain. The Microbot's performance on hard clay terrain with large rocks will be different than its performance on very fine, flour-like sand. Also, the conditions inside the lava tubes will have a significant impact on the mobility system performance. These issues should be carefully considered during the design of the mobility system.
Chapter 4: Feasibility Analysis
78
CHAPTER
5
This thesis presents a feasibility analysis of the Microbot power and mobility systems. To evaluate the feasibility of a fuel cell power system, an experimental prototype was developed. The device was used to power a Microbot prototype. Analysis of the power system performance and a comparison with the predicted Microbot power system requirements suggests that fuel cells are a feasible power generation option for the
Microbot concept.
The feasibility of the hopping mobility system was also examined. The robotterrain interactions were explored analytically and through laboratory experiments.
Dynamic simulations were created to test the Microbot mobility approach in virtual Mars terrains, including lava tube caves. The results of this work suggest that the hopping mobility system will have some success traversing the Martian terrain. The Microbot will not become totally submerged in the soft Martian sand and the hopping mobility system will allow the Microbots to overcome certain rock formation obstacles.
Finally, the feasibility of the Microbot was examined through a power and mobility system level study. A mass-energy model was proposed and the model performance was evaluated for a Mars reference mission. The results of the study show that the Microbot can successfully complete the reference mission of over 2000 hops on
Mars with the design specifications presented in Table 1. The study also highlights the
Chapter 5: Conclusions and Recommendations 79
important coupling of the Microbot energy consumption, mission duration, system mass, and mobility system size.
A number of technologies need to be developed further to progress this research and to make the Microbot concept a reality. First, a miniaturized version of the fuel cell power system should be developed. Much of the technology required to fit the power system in the Microbot already exist, however the components need to be integrated and made lightweight. A miniature hydrogen storage system appropriate for the Microbot will also need to be developed. This is an area of intense research and these technologies should be available in the next 10 years.
The mobility system research also requires further development. The interactions between a large number of Microbots and between the Microbots and the terrain are not fully understood at this point in time. A realistic dynamic simulation that investigates the performance of hundreds of Microbots would greatly aid in the mobility system design.
Several technical design elements of the mobility system also require further development, including the directionality mechanism and the DEA device. These are areas of ongoing research. Much progress is still required to create a device capable of generating the thousands of hops required to complete the Mars reference mission.
Assuming that innovations continue to take place in the area of DEA research, the hopping mobility system should be completed in approximately 10 years.
In the long term, the development of a fully integrated Microbot prototype will require the efforts of researchers in many fields. Ultra-low power sensors and navigation electronics have to be miniaturized and integrated into the Microbot system. The development of a device to deploy Microbots on Mars is also required. This device could
Chapter 5: Conclusions and Recommendations 80
be a landing craft, a Microbot transportation container for a conventional rover, or an aerial deployment device like a balloon or an aircraft. The research in these areas has just begun, therefore these technologies should be developed in at least 20 years.
5.3 Outlook
The Microbot concept is a unique and innovative space exploration system. The redundant robots concept makes the system robust to failure. The hopping mobility system performance is promising for subterranean planetary exploration. While some technological challenges currently exist, the Microbot components are of great interest to researchers. Much research is currently being conducted in the areas of fuel cells, DEAs, and MEMS sensors. These technologies will hopefully advance greatly in the near future, enabling the development of the Microbot.
The NASA Institute for Advanced Concepts (NIAC), the institution that funded this research, was interested in the development of technologies that will be achievable in
"10 to 40 years in the future [44]." It is believed that in this period of time, the required technologies for the Microbot will become available. Considering the current rate of research advancement in the Microbot-related fields, the system should be ready for deployment on Mars as part of a future exploration mission.
Chapter 5: Conclusions and Recommendations
81
81
1
2
Angstrom Power, 2006, V60 Fuel Cell Module. Available: http://www.angstrompower.com/productsv60.html
F. Barbir, PEM Fuel Cells: Theory and Practice. London: Elsevier Academic
Press, 2005.
M. G. Bekker, Theory ofLand Locomotion: University of Michigan Press, 1956. 3
4
5
6
7
8
9
P.J. Boston, "Extraterrestrial Caves," Encyclopedia of Cave and Karst Science.
London: Fitzroy-Dearborn Publishers, 2003. Pp. 355-358.
P.J. Boston, R. D. Frederick, S. M. Welch, J. Werker, T. R. Meyer, B.
Sprungman, V. Hildreth-Werker, and S. L. Thompson, "Extraterrestrial subsurface technology test bed: Human use and scientific value of Martian caves," in Space Tech. & Applic. Forum, College Park, MD, 2003.
B. R. Burg, S. Dubowsky, J. H. Lienhard, and D. Poulikakos, "Thermal Control
Architecture for a Planetary and Lunar Surface Exploration Micro-Robot," be published in The Space Technology and Applications Int.l Forum, Melville, NY,
2007.
K. A. Burke, "Small Portable PEM Fuel Cell Systems for NASA Exploration
Missions," in Proc. of3rd AIAA International Energy, San Francisco, CA, 2005.
K. A. Burke, "Fuel Cells for Space Science Applications," in First International
Energy Conversion Engineering Conference Portsmouth, Virginia,: AIAA, 2003.
M. Buschmann, S. Winkler, T. Kordes, H. Schulz, and P. V6rsmann,
"Development of a fully autonomous Micro Aerial Vehicle (MAV) for Ground
Traffic Surveillance," in 10th IFAC Symposium on Control in Transportation
Systems, Tokyo, Japan, 2003.
10 R. Cross, "Measurments of the horizontal coefficient of restitution for a superball and a tennis ball", American Journal ofPhysics 70, 482-489, 2002.
11 A. C. Dillon and M. J. Heben, "Hydrogen storage using carbon adsorbents: past, present and future," Applied Physics A: Materials Science & Processing, vol. 72,
pp. 133-142, 2001.
12 S. Dubowsky, K. lagnemma, and P. J. Boston, "Microbots for Large-Scale
Planetary Surface and Subsurface Exploration," Phase I report, NASA Institute for Advanced Concepts, 2004.
References 82
13 S. Dubowsky, K. Iagnemma, S. Liberatore, D. M. Lambeth, J. S. Plante, and P. J.
Boston, "A Concept Mission: Microbots for Large-Scale Planetary Surface and
Subsurface Exploration," in The Space Technology and Applications
14 Dust Networks Inc ((2006, August 10). SmartMesh-XT M2030. Available: http://www.dustnetworks.com/docs/M2030.pdf
15 EMCO High Voltage Power Supplies (2006, August 10). http://www.emcohighvoltage.com/
Q
Series, Available:
16 G. J. Fischer and B. Spletzer, "Long range hopping mobility platform," in SPIE
Unmanned Ground Vehicle Technology Conference, Orlando, FL, 2003, pp. 83-
92.
17 P. Fiorini, C. Cosma, and M. Confente, "Localization and sensing for hopping
robots," Autonomous Robots, vol. 18, pp. 185-200, 2005.
18 P. Fiorini, S. Hayati, M. Heverly, and J. Gensler, "A Hopping Robot for Planetary
Exploration," in IEEE Aerospace Conf, Snowmass, CO, 1999.
19 P. Fiorini, and J. Burdick, "The Development of Hopping Capabilities for Small
Robots," Autonomous Robots, vol. 14, March 2003, pp. 239-254.
20 Foster-Miller Inc., "Kugel report," 2006. Used with permission.
21 S. Fraser, M. Monsberger, and V. Hacker, "Fuel Cell Power System Options for
Mars Rovers," in 2nd International Conference on Green Propellants for Space
Propulsion Chia Laguna, Sardinia, Italy: ESA, 2004.
22 G. R. Frysinger (2007, June). http://www.galenfrysinger.com/idaho.htm
Travel Photos, Idaho, Available:
23 M. Golombek and D. Rapp, "Size-frequency distributions of rocks on Mars and
Earth analog sites: Implications for future landed missions," Journal of
Geophysical Research, vol. 102, pp. 4117-4129, February 25 1997.
24 R. M. Haberle, C. P. McKay, J. Schaeffer, N. A. Cabrol, E. A. Grin, A. P. Zent, and R. Quinn, "On the possibility of liquid water on present-day Mars," J.
Geophys. Res., vol. 106, pp. 23,317-23,326, 2001.
25 E. Hale,. N. Sehara, J. W. Burdick, and P. Fiorini, "A Minimally Actuated
Hopping Rover for Exploration of Celestial Bodies," in IEEE International
Conference on Robotics and Automation, San Fransisco, CA, 2000, pp. 420-427.
References
83
26 K. Jagnemma and S. Dubowsky, Mobile Robots in Rough Terrain: Estimation,
Motion Planning, and Control with application to Planetary Rovers vol. 12:
Springer, 2004.
27 S. B. Kesner, J. S. Plante, P. J. Boston, T. Fabian, and S. Dubowsky, "Mobility and Power Feasibility of a Microbot Team System for Extraterrestrial Cave
Exploration," in 2007 IEEE International Conference on Robotics and
Automation, Rome, Italy: IEEE, 2007.
28 S. B. Kesner, J. S. Plante, P. J. Boston, and S. Dubowsky, "A Hopping Mobility
Concept for a Rough Terrain Search and Rescue Robot," in International
Conference on Climbing and Walking Robots, Singapore: CLAWAR Association,
2007.
29 R. Kombluh, R. Pelrine, and J. Joseph, "Elastomeric Dielectric Artificial Muscle
Actuators for Small Robots," in The Materials Research Society Symposium,
Boston, MA, 2000.
30 J. Larminie and A. Dicks, Fuel Cells Systems Explained, 2nd ed. Chichester,
England: Wiley, 2003.
31 Maxim Integrated Products (2006, August 8). MAX1705 1 to 3 Cell, High
Current, Low-Noise, Step-Up DC-DC Converters with Linear Regulators.
Available: http://www.maxim-ic.com/quick-view2.cfm/qvpk/1720
32 H. H. Meinel, "Commercial applications of millimeter waves: history, present status, and future trends," IEEE Transactions on Microwave Theory and
Techniques, vol 43, pp. 1639-1653, July 1995.
33 MER Analysts Notebook (2007, June), "Endurance Crater DTM, ASCII Format",
Available: http://anserverl.eprsl.wustl.edu/anteam/merb/merosu.htm
34 MSC Software, "ADAMS/View Help," Version 2005 2.0., accessed July, 2007.
35 Y. Nakamura, S. Shimoda, and S. Shoji, "Mobility of a microgravity rover using internal electro-magnetic levitation," in IEEE International Conference on
Intelligent Robots and Systems, Takamatsu, 2000, pp. 1639-1645.
36 NASA (2003, Sept.). Astrobiology Roadmap (Final Version). Available: http://astrobiology.arc.nasa.gov/roadmap.
37 NASA (2007, June.). Mars Exploration Rover Mission, Images Gallery.
Available: http://marsrovers.nasa.gov/gallery/images.html
38 R. O'Hayre, S. Cha, W. Colella, W., and F. B. Prinz., Fuel Cell Fundamentals,
New York, NY: Wiley, 2005.
References 84
39 J.S. Plante, "Dielectric Elastromer Actuators for Binary Robotics and
Mechatmics," PhD thesis, Dept. of Mech. Eng., Massachusetts Institute of
Technology, Cambridge, MA, 2006.
40 J. S. Plante, M. Santer, S. Pellegrino, and S. Dubowsky, "Compliant Bistable
Dielectric Elastomer Actuators for Binary Mechatronic Systems," in Proc. of the
ASME Conference on Mechanisms and Robotics, Long Beach, CA, 2005.
41 Pico Electronics (2006, August 10). Series VV Ultra-Miniature 6,000 to 10,000
VDC Outputs Available: http://www.picoelectronics.com/dcdclow/pe66_
67hv.htm
42 K. Prater, "The Renaissance of the Solid Polymer Fuel Cell," Journal of Power
Sources, vol. 29, pp. 239-250, January 1990.
43 M. H. Raibert, Legged Robots That Balance, Cambridge, Mass: MIT Press, 1986.
44 NASA Institute for Advanced Concepts (2007, June 10). NIAC Home page,
Avaliable: http://www.niac.usra.edu/
45 S. Shimoda, T. Kubota, and I. Nakatani, "New mechanism of attitude control for hopping robot," in IEEE International Conference on Intelligent Robots and
Systems, Lausanne, Switzerland, 2002, pp. 2631-2636.
46 S. Shimoda, A. Wingart, K. Takahashi, T. Kubota, and I. Nakatani, "Microgravity
Hopping Robot with Controlled Hopping and Landing Capability," in IEEE/RSJ
International Conference on Intelligent Robots and Systems, Las Vegas, NV,
United States, 2003, pp. 2571-2576.
47 P. H. Smith, J. F. Bell, III, N. T. Bridges, D. T. Britt, L. Gaddis, F. Gliem, R.
Greeley, H. U. Keller, K. E. Herkenhoff, S. Hviid, R. Jaumann, J. R. Johnson, R.
L. Kirk, M. Lemmon, J. N. Maki, M. C. Malin, S. L. Murchie, J. Oberst, T. J.
Parker, R. J. Reid, P. Rueffer, R. Sablotny, L. A. Soderblom, C. Stoker, R.
Sullivan, N. Thomas, M. G. Tomasko, W. Ward, and E. Wegryn, "Results from the Mars Pathfinder Camera," Science, vol. 278, pp. 1758-1765, December 5,
1997.
48 S. A. Stoeter, P. E. Rybski, M. Gini, and N. Papanikolopoulos, "Autonomous stair-hopping with scout robots," in IEEE/RSJ International Conference on
Intelligent Robots and Systems, Lausanne, Switzerland, 2002, pp. 721-726.
49 H. Tsukagoshi, M. Sasaki, A. Kitagawa, and T. Tanaka, "Jumping robot for rescue operation with excellent traverse ability," in IEEE International
Conference on Advanced Robotics, Seattle, WA, United States, 2005, pp. 841-
848.
References 85
50 H. Tsukagoshi, S. Shimoda, K. Iizuka, T. Kubota, and I. Nakatani, " A Study of
Locomotion Mechanism based on Gravitational Environment," in IEEE
International Conference on Intelligent Robotics and Systems, Sendal, Japan,
2004.
51 H. Tsukagoshi, M. Sasaki, A. Kitagawa, and T. Tanaka, "Design of a higher jumping rescue robot with the optimized pneumatic drive," in IEEE International
Conference on Robotics and Automation, Barcelona, Spain, 2005, pp. 1276-1283.
52 US Department of Energy (2006, August 10). HFCIT Fuel Cells: Types of Fuel
Cells Available: http://wwwl.eere.energy.gov/hydrogenandfuelcells/fuelcells/ fc-types.html
53 US Department of Energy (2006, August 10). HFCIT Hydrogen Storage: Metal
Hydrides Available: http://www.eere.energy.gov/hydrogenandfuelcells/storage/ metalhydrides.html
54 J. Vogan, "Development of Dialectric Elastomer Actuators for MRI
Applications," M.S. Thesis, Dept. of Mech. Eng., Massachusetts Institute of
Technology, Cambridge, MA, 2004.
55 B.A. Warneke, and K.S.J. Pister, "An ultra-low energy microcontroller for Smart
Dust wireless sensor networks," in IEEE Int. Solid State Circuit Conf, San
Francisco, CA, USA, 2004.
56 A. N. Wilhelm, B. W. Surgenor, and J. G. Pharoah, "Design and evaluation of a micro-fuel-cell-based power system for a mobile robot," IEEE-ASME
Transactions on Mechatronics, vol. 11, pp. 471-476, Aug 2006.
57 J. Y. Wong, The Theory of Ground Vehicles, 3rd ed. New York: Wiley, 2001.
58 J. J. Zakrajsek, D. B. McKissock, J. M. Woytach, J. F. Zakrajsek, F. B. Oswald,
K. J. McEntire, G. M. Hill, P. Abel, D. J. Eichenberg, and T. W. Goodnight,
"Exploration Rover Concepts and Development Challenges," in 1st Space
Exploration Conference: Continuing the Voyage of Discovery, Orlando, Florida:
AIAA, 2005.
86
86
APPENDIX
A
A.1 Introduction
The fuel cell experiment was designed and constructed for the purpose of demonstrating the possibility of a fuel cell power system for trickle-charging a battery and powering the Microbot DEA. To accomplish this, the following components were used:
9 Large tank of industrial quality Hydrogen Gas, Airgas part # HY200.
9 Precision gas regulator, Part # COA 4021361-350, A 402 series Regulator.
* Hydrogen gas humidifier composed of a beaker, a rubber plug, tubing, and distilled water (See Figure 8).
* 3 Air-breathing PEM fuel cells provided Dr. Tibor Fabian of Stanford
University.
0 Low voltage regulator set to an output voltage of 5 volts, part #
MAX1703EVKIT.
* PIC microcontroller, part # PIC 12F5 10.
* Single P-channel MOSFET, part # TPS1 100.
* Single cell lithium-ion battery charger, part # MAX1926EVKIT.
* Low voltage regulator set to 4.3V out, part # MAX1703EVKIT.
* EMCO High voltage converter, part # Q101.
Appendix A: Details of The Fuel Cell Experiment 87
"
Custom Microbot DEA.
* Resistors and capacitors for measuring, filtering, and discharging the
DEA. Including: o 1 00k to measure the DEA current.
o 0.75K to measure the fuel cell.
The first step in running the experiment is to warm up the fuel cells. Initially the cells are purged with a constant flow of hydrogen gas until a voltage of 0.7 volts is seen in each cell. After that, the exit port is blocked off, the gas regulator is set to
5 psi, and the cells are run in dead-end mode. The PEM electrolyte must be at a warm temperature to operate efficiently, so the cells are shorted individually for a period of time. They are shorted until the stack open circuit voltage is over 2.7 volts.
The hydrogen fuel is humidified to prevent the electrolyte membrane from drying out. A bubble-humidifier is used. This device uses gas pressure to force the hydrogen gas bubbles through the container of distilled before they enter the fuel cells. The gas enters the humidifier through a tube that carries it to the bottom of the container where it rises through the water and exits through another tube in the rubber stopper. See Figure
8.
Section 2.2.2 outlines the experimental procedure. The measurements were taken primarily with a voltage-sensing oscilloscope. The fuel flow rate was approximated by counting the number of bubbles that flow through the humidifier during a certain time
Appendix A: Details of The Fuel Cell Experiment
88
period. The volume of the bubbles was determined by averaging measurements taken from images such as the one in Figure 33. Flow rates of 3.8 micrograms/sec without charging the battery and 6.4 micrograms/sec while charging the battery were recorded.
Figure 33: The setup used to measure the bubble volume and approximate the fuel flow rate.
The fuel cell voltage and current were measured with the oscilloscope. The voltage across the stack and across a 0.75Q current-sensing resistor on the cathode side of the cells was measured. The voltage across the DEA was measured via the 1 gn resistor used to discharge the DEA after each actuation. The current was measured via a 100 k 9 resistor on the ground side of the high voltage converter.
89
89
APPENDIX
B
The mobility simulation was conducted in MSC's ADAMS/view simulation software. The terrain was imported as a parasolid (extension: .x_t) generated by
Solidworks. See Appendix C for more technical details of the terrain generation process.
The Microbot mobility control system was implemented in Mathwork's Simulink software in conjunction with Matlab.
After the terrain was imported, the rock piles were constrained to the greater terrain body (composed of the hill, plain, and lava tube cave) with a fixed constraint. The microbot was then defined. The Microbot ADAMS model parameters are the following:
90 Appendix B: Mobility Simulation Details
Table 13: The Microbot model parameters.
Microbot Model Parameter
Mass
Diameter
Value
0.1 kg
0.1 m
Moment of Inertia
Center of Mass
Hoping Force
0 1 0
0 0 1
(0,0,0) relative to the body_
General Force applied at the Microbot Center of Mass
After the Microbot was defined, the appropriate contact models were created between the Microbot, the terrain, and the rock piles. The contact model parameters are discussed in section 4.2.1.
Once all of the ADAMS/view model components were completed, the Simulink control system was created. First, the ADAMS/control plug-in was added to the model and the appropriate state variables were defined in ADAMS/view. Once these steps were taken, the Matlab and ADAMS control files were generated. These files were then loaded into Matlab, thus generating an ADAMS subsystem block that was inserted into a
Simulink control system. Figure 34 shows the control system design.
91
91
.r., oely m.dgcal
Or]x holez
"p3 hopA.
hop hCPX ftpy h. p hopy
Pz
T_..
zWC mitial Value
0 gt
NIATLAS
.cdor
MATLIAB Fwn lignam8 sig nats sign allO sign011' srIn12 ignal1 sign|15
+ signs14 --
~ sigr~I ponz
To Woftac birnay
Figure 34: The ADAMS simulation Simulink control system.
The Simulink control system allows ADAMS to communicate with Matlab. The
"adamssub" block transfers the ADAMS/view simulation output variables to the control system and returns the Simulink-generated control signals as inputs. The ADAMS output variables are the position of the Microbot, the end goal, and an intermediate destination.
In the case of the reference mission simulation, the intermediate goal was the first skylight and the end goal was the far end of the lava tube cave. The Simulink model uses
Appendix B: Mobility Simulation Details
92
a custom Matlab Function block to calculate the direction and start time of the next hop and to generate the control signal. The input variables to the ADAMS model are the force impulses that cause the Microbot to hop.
The simulation was run through the Simulink software. The length of the simulation and the time step rate were set through the Simulink model. To automate the process, a Matlab m-file was created to allow for a number of simulations to be run in a series.
Both the Simulink control system and the ADAMS model can be modified through the m-file. The Simulink model can be changed directly to vary any of the control signal parameters, including the hop height, hop directions, and simulation duration. To modify the ADAMS model, the m-file must load and modify the ADAMS adm-file. This file contains all of the ADAMS/view simulation parameters. By editing the adm-file, the user can change the Microbot start and goal positions and the contact model parameters.
Appendix B: Mobility Simulation Details 93
APPENDIX
C
C.1 Introduction
The simulation terrain was created over a number of design iterations. The purpose of the terrain is to simulate key aspects of the Mars reference mission, including a rough and irregular surface, a lava tube cave with skylights, and rubble obstructions where the lava tubes have collapsed to form skylights.
C.2 Matlab
The first component of the terrain that was constructed was the surface. The surface is composed of a hill created by inverting a 3D image of the Martian crater
Endurance and a larger flat area The Endurance image was recorded by the stereo cameras on the Mars Exploration Rover (MER) Opportunity [33]. Roughness was then superimposed over the surface of the terrain. The equation to describe this roughness was derived from rock size distribution data from a Viking Landing site [23]. The equation used was the following:
N(D) = L-sD
Appendix C: Generating the Simulation Terrain
(C. 1)
94
where D is a rock diameter, N is the number of rocks per square meter equal to or greater than D, L is the total number of rocks per square meter, equal to 6.84 in this case, and s is equal to 8.3 for the landing site. This equation was solved for D given a random value of N. The rock diameter values were used to set the relative height of the facets in the terrain surface mesh.
The terrain was generated as a Matlab surface matrix and then converted to a stlfile (stereo lithography file). The stl-file format was selected because it can be imported into the solid modeling tool Solidworks.
Solidworks was selected as the solid modeling software to create the terrain. The stl-file surface generated in Matlab was imported into Solidworks, and a solid part was extruded up to the surface. A semicircular cut was then made into one of the new part's faces up to the hill. This cut created the lava tube cave. The cut was made close enough to the surface to create skylight openings where the surface topology was low enough.
The rock piles were also created in Solidworks. Cubes with a side length of 10,
20, and 40 cm were created. Each of the cubes was then modified with facets, tapers, and other features. The rocks were then assembled into larger piles of rocks with an approximately random distribution. The rocks in each pile were then joined together to make a single bodied object.
C.4 Assembly and Conversion
The final step in the terrain generation process was the assembly and conversion of the terrain. A Solidworks assembly file was created. The main surface and lava tube
Appendix C: Generating the Simulation Terrain
95
cave part as well as the rock piles were imported. The rock piles were placed underneath each of the lava tube cave openings.
After all of the rock piles were orientated in the correct locations and the terrain was finished, it was converted into a parasolid x t-file. Parasolid is a format that allows the terrain to be imported into the ADAMS/view model.
96
96