Cool Robot
Mechanical Design of a SolarPowered Antarctic Robot
Alex Price
Advisor: Dr. Laura Ray
Thayer School of Engineering at Dartmouth College
Project Goals

Traverse the Antarctic south polar plateau
autonomously on renewable energy
 Relatively cheap (about $20,000)
 Travel 500 kilometers in 2 weeks
 Easy to handle, transport, and maintain
–
–
–
–
As lightweight as possible (also for energy reasons)
Small enough to fit inside the Twin Otter aircraft.
Easily assembled and tested after delivery
Scientific instruments easily added and integrated
2
Antarctic Plateau

Large central flat plateau
–
–
–
–
–

High altitude (2800 meters)
Cold (-20° to -40° C in summer)
Dry and sunny, but windy
Firm, clean snow
Flat, but with wind-sculpted
“sastrugi” snow drifts
Possible Robot Missions
– Automated distributed sensing


Magnetometers
Ionosphere studies
– Ground-penetrating Radar
– Traverse team support
– Ecological Studies
Sastrugi
3
Specifications and Solutions

Specifications:
– Average Speed of 0.4 m/s, top speed at least twice that
– Maximum dimensions to fit in Otter:



1.5 m long
1.2 m wide
1.2 m tall
– Less than 75 kg empty ; 15 kg payload capacity.
– Maximum ground pressure of 3 psi

Design to achieve those goals:
–
–
–
–
Specialized lightweight construction
Optimized dimensions
Careful component selection (tires, bearings, etc.)
Custom wheels, hubs, and drive train components
4
Overall Robot Design



Solar panels attached over chassis
and wheels by support arms
Tube on top of chassis box may be required to
support center of top panel
Insulation is likely not required.
5
Solar Power in the Antarctic

In summer, sun never sets, but is
always at a low angle
 Sun is brighter in high, dry climate

Significant reflected light from
snowfield
– Proportional to sun azimuth
– Snow albedo of as high as 0.95


Diffuse component of insolation as
large as 100 W/m2 from atmospheric
scattering
Sunny day insolation fairly constant,
but scattering and cloud cover varies
with the time of year.
30
Azimuth (deg. above horizon)
– As bright as 1200 W/m2 on a clear day
– Few cloudy days in the central plateau
Sun Azimuth vs Day of Year
at 85° South Longitude
25
20
15
10
5
0
0
60
120
180
240
Day of Year (1 = Jan. 1st)
Max Azimuth
300
360
Min Azimuth
Variation in azimuth
between max and min
decreases to zero at 90°,
at the pole.
6
Solar Power in the Antarctic
Top (direct sun only)
34%
Back (in shadow)
Front
128%
11%
Sides
34%
(reflected light only)
Available Power in Average Summer Sun:
1000 W/m2 of solar power available on an average sunny day
Sun azimuth angle 20° from horizon (average for November-February)
Robot facing front towards sun (worst case) ; Snow albedo 90%
Panel capacities are based on nominal 1-sun (1000 W/m2) input:
100% = 200 W/m2 energy output (20% efficient cell in direct sun)
7
Scaling Capability
Performance of Different Robot Size Options (in average sun)
140%
Empty Mass (% of 75 kg goal)
120%
Power Available (% of full power)
100%
Maximum Speed (% of 1 m/s)
80%
60%
40%
20%

Robot Configuration
al
l
Sm
8x
5
9x
5
9x
6
ed
iu
m
M
La
rg
e
0%
Notes on configurations:
Large = 11x9x8 cells (9x9 on top)
Med. = 10x9x6 cells (9x8 on top)
9x6, 9x5, and 8x5 sizes are
"cubic" with that size panel on
each of the 5 sides.
Small = 7x7x4 cells (7x4 on top),
also cubic, with only 2 motors.
Design can be scaled well to a variety of
sizes for different mission goals.
8
Tire Selection
ATV tires
Custom cut tire
Russian Snow Bug tire
Apollo 17 rover
mesh wheel

Roleez ballon tire
Mars Rover solid wheel
Ideal tire would be lightweight and would have good
traction, low ground pressure, and low rolling
resistance; but no such tires are available within budget.
9
Tire Selection

Best tire of available selection was Carlisle’s
16x6-8 knobby ATV tire
– About 6.5 pounds, very stiff, good tread pattern
10
Wheel Design
ITP aluminum

Carlisle steel standard
1st design iteration
Commercially available wheel options are not suitable.
– Aluminum racing wheels are all too large
– Available 8”x5.5” wheels are too heavy (> 2.3 kg)
– Require the use of heavy bolts and hubs

Thus, a custom wheel had to be designed to meet the
requirements of the design
11
Wheel Design

Factor of Safety of 3 against static failure in
worst-case loading
 Factor of Safety of at least 2 against fatigue failure
in worst-case driving conditions
 Only 0.9 kg, and uses smaller bolts & hub
 Tubeless if 2 halves are sealed
12
Hub Design

Standard 4-inch bolt circle
 Welds to drive shaft, bolts to wheel tabs
 Factor of safety of at least 2.5 against fatigue
failure in worst-case loading
13
Assembled Wheels

Wheel + hub + nuts and bolts = 1.1 kg
– Far better than the commercially available 3+ kg

Total assembly (with tire and covers) = 4 kg
 Total weight savings on robot = 8 to 9 kg
14
Drive Train
Option 1: Cantilevered
support tube with press-fit
bearing, minimizes loads
on gearbox.
Option 2: Bearing pair to
carry load, motor mounted
loosely so bearings will
support the bending loads.

Very efficient motor and gearbox
 Custom hollow aluminum shaft and supports
15
Integration and Assembly

Heaviest components mounted in the center
 Motors, controllers, power electronics, and scientific
instruments mounted symmetrically on chassis
16
Future Plans and Goals

Complete Design and Test Components
–
–
–
–
Wheels and Hubs NC machined
Drive Train design completion
Assemble and test drive train
Assemble and test solar panels

July - Chassis operational
on batteries
 August - Solar power systems
tested and operational
 September - Robot operational
on solar power
 Next year - Testing in Greenland and in Antarctica!
17
Conclusions

Design has been
optimized within the
strict parameters
 Robot should easily
meet the mission goals
 Future versions could
be lighter and faster.

Autonomous navigation at the south pole is a daunting
task, but we are well on our way to achieving that goal.
 Building a robot is a lot of work, but has been and will
continue to be a great experience.
18
Acknowledgements












Laura Ray
Alex Streeter
ENGS 190/290 group
Guido Gravenkötter
Gunnar Hamann
Mike Ibey
Kevin Baron
Pete Fontaine
Leonard Parker
Paula Berg
Cathy Follensbee









Jim Lever
Dan Denton
CRREL
Marc Lessard
Gus Moore ‘99
Michael at Wilson Tire
Don Kishi at Carlisle Tire
National Science Foundation
Everyone at Thayer School
who has made this possible
Full reference and bibliography information is included in the report.
19