International Journal of Engineering Trends and Technology (IJETT) – Volume 26 Number 2- August 2015
Implementation of Soft Computing Algorithm for Path
Finding Robot Using Soft Core Processor
Prof. Vilas V. Deotare#1, Dheeraj D. Jain*2, Prof. Dinesh V. Padole#3
# Head of Department, Electronics & Telecommunication, SIT, Lonavala, Maharashtra, India
*
PG students, Electronics & Telecommunication, SIT, Lonavala, Maharashtra, India
# Professor, Electronics & Telecommunication, GHRCE, Nagpur Maharashtra, India
Abstract— Path finding of robots can be achieved
by many algorithms and many optimization
techniques applied for different parameters and
their architectures, this paper explains how some
parameters affect performance of path finding
robot and after using hardware software co-design
how computational time required can be optimized.
After applying optimization technique optimum
values can be obtained. The optimization
techniques like genetic algorithm (GA) gives
optimized values of parameters used in fitness
function. The whole application is designed using a
soft processor which enhances the performance of
those attributes which are not beneficial in
hardware as well as software design. The area
required for this design is less than that of
hardware and which improves the performance.
Keywords: Optimization technique, Soft core
processors, Path finding robot, Hardware software
co-design
I.
INTRODUCTION
Designing efficient autonomous mobile
robot controllers is a difficult and time consuming
process which requires wasting resource sand
efforts. Commonly, the robot navigation is based
on continuous interaction between the robot and the
environment. Providing the required degree of
flexibility and maintaining robustness is one of the
relevant challenges when designing models for
robot controllers because of the mathematical
complexity involved in the necessity of taking into
account any possible emerging change in the
environment. Robotic their development and its
research is important subject in automation
industries. Robotics is trending all over the world
because of its applications which are there in each
and every field. In case of application specific areas
the automation is essential one. The main objective
of researchers is to optimize area, computational
time, accuracy. There are many ways to achieve the
above mentioned objectives. These objectives can
be using different architectures, algorithms,
hardware and software. The hardware used for this
application is either a FPGA or a microcontroller.
The main disadvantage of using simply hardware to
solve problem is it requires more area and another
one is cost of hardware is more than that of the
software. Some researchers have used just software
ISSN: 2231-5381
part to solve the same problem but the ambiguities
while using software is its speed is less and
computational time is more as compared to
hardware though the cost is less and its actual
implementation is difficult.
Artificial Intelligence is one of the
solutions to give optimized path and obtain
required results but the complexity increases and in
dynamic environment the computational time may
increase. So, to achieve the advantages of both
hardware and software it’s better to go with
hardware software co-design. In the combination of
these two, the application parameter can be
optimized and will give balanced results. This
paper explains how hardware- software co-design
can be used for path finding application in static
environment and it shows optimized and balanced
results.
There are many optimization algorithms
one can use Multi attribute Decision Making
(MADM) when there is single objective and having
multiple attributes and multiple alternatives. But
Multi objective Decision Making (MODM)
techniques are used to achieve multiple objectives
from one application. For particular problem
already so many MODM techniques have been
used along with their improved versions. But in this
paper we have used genetic algorithm (GA)
optimization algorithm which has shown better
results for maximum functions and applications as
proved.
II.
RELATED WORK
For the application of path finding mobile robot
many researchers have done experiment with
different hardware, software and algorithms.The
path finding system can be implemented using
FPGA along with artificial intelligence and HPOPSO algorithm witch gives better results as
compared to previous implementations [1].
In the next paper the 2-D cell architecture has been
presented for the Euclidean distance Transform
(EDT) and Nearest Neighbour Transform (NNT)
for operating at high speed [2]. In this technique the
image of arena is captured by camera mounted on
top after that the image is converted into binary
image and then using EDT and NNT shortest path
http://www.ijettjournal.org
Page 78
International Journal of Engineering Trends and Technology (IJETT) – Volume 26 Number 2- August 2015
is calculated. Optimization is the field of research
where each and every attribute of application can
be optimized depending on its limits.
The genetic FPGA implementation for
path planning of autonomous mobile robot gives
optimized results and Genetic algorithm (GA)
works perfectly even if an environment is unknown
[3].
A Modified Particle Swarm Optimizer
(MPSO) algorithm is used to find optimal path for
the mobile robot in working environment with
obstacles is capable of effectively guiding a robot
moving from start position to the goal position in
complex environment and find optimum/shortest
path without colliding any obstacles in the
environment. [4].
However, it takes a lot of time to get the
solution and it is not too easy to obtain the optimal
path every time. It is also difficult to apply to the
complex and big size maps so ACO algorithm are
changed to converge into the optimal solution
rapidly when a certain number of iterations have
been reached [5].
III.
B. Description of implementation
The whole system behaviour is described
here. The Spartan kit is connected to computer
through RS-232 serial cable and USB cable which
become medium of communication between two
devices in which, the code is implemented on
hardware using USB cable and further
communication is done by RS-232 cable.
The baud rate affects the serial communication.
Here, in this particular application the baud rate
should be in between 2400 to 11500, but we have
kept it as 9600 as per our hardware requirements.
After setting up an environment all VHDL files are
run in Xilinx 8.01i.
The path finding algorithm is implemented
on hardware and the robot communicates with
hardware finds a path according to that after
obstacle is detected at simulator side the simulator
sends the sensor values and obstacle detected
message after which the Pico blaze processor runs
an algorithm which sends and solution to further
process.
IMPLEMENTATION
A. FPGA concepts
FPGAs are field-programmable logic devices
which contains matrix of Configurable Logic
Blocks (CLBs) interconnected by an array of
routing resources implemented in CMOS
technology. CLB features depend on both
producers and family devices; however, they are
typically small tables with 4,5 or 6-bitinputs,
namely Look up Tables(LUTs),D flip-flops and
several multiplexers, allowing the truth value table
of basic Boolean functions to be implemented in
hardware. In order to implement complex circuits,
CLBs are connected by a programmable network of
connection and switching blocks. The connection
block allows logic block inputs and outputs to be
assigned to horizontal or vertical tracks. The
switching block allows a signal on a track to
connect to another track. The connections in the
switching and connection blocks are made by
programmable points. Commonly, a programmable
point consists of a pass transistor controlled by a
static random access memory cell(SR)to hold the
user defined configuration values depicts a planar
switching box topology in which a wire in track
number connects only to wires in track number in
adjacent channel segments. Modern FPGA devices
contain embedded DSP blocks, RAM blocks,
dedicated processors and digital clock managers
allowing the implementation of more complex
designs.
Fig.3.1 Program flow of path finding robot
ISSN: 2231-5381
http://www.ijettjournal.org
Page 79
International Journal of Engineering Trends and Technology (IJETT) – Volume 26 Number 2- August 2015
divided to other parameters. The fitness function is
formed according to parameters extracted
C. Picoblaze
Pico Blaze is the designation of a series of three
free soft processor cores from Xilinx for use in
their FPGA and CPLD products. Here Pico blaze
soft processor is been used as embedded processor
to run algorithm. Its architecture and features are
shown below:
(1)
Where,
- Speed of robot in vertical direction
– Speed of robot while taking turn
2) Parameter dependencies:
As these two parameters are dependent on other
parameters therefore their dependencies is being
calculated and equations are formed
(2)
(3)
–Speed of robot in RPM
– Distance value i.e. distance between robot and
arena
Fig.3.2 Pico blaze processor architecture
D. Eyesim simulator:
The Eye Bot simulator EyeSim is a multiple mobile
robot simulator that allows experimenting with the
same unchanged EyeBot programs that run on the
real robots. EyeSim includes simulation of the
robot's driving actuators (differential steering), as
well as all robot sensors, including:
On-board vision
images)
Infra-red sensors
Bumpers
(synthetic
After substituting equation (2) and (3) in equation
(1) we get final fitness function. The limits are
being drawn after so many iteration and trial and
error method. All parameters are converted in equal
units like RPM is converted to RPS.
Fitness Function becomes,
F(x)
=
2 * 3600(4)
generated
Limits,
100 ≤
0<
≤ 500
≤1
A 3D scene representation of the environment and
all robots in it is being shown, together with a list
of active robots. Clicking on a robot from this list
will identify the robot in the environment and also
open a window to show its control panel
(equivalent to the LCD display and buttons on the
EyeBot controller) for communicating via the
robot's user interface.
E.Genetic algorithm and its implementation
1)Parameter extraction and equation formation
In this path finding robot application, parameters
have been extracted and optimization technique is
implemented. The extracted parameters and fitness
function of parameters is shown in below equation.
The parameters which are beneficial for the
performance/speed of robot are multiplied with
other parameters and which are non-beneficial are
Fig.3.2 GA based flowchart of proposed technique.
ISSN: 2231-5381
http://www.ijettjournal.org
Page 80
International Journal of Engineering Trends and Technology (IJETT) – Volume 26 Number 2- August 2015
Fig. 4.1 Graph of fitness function
Now, here the comparison of fitness value obtained
from soft computing algorithm i.e. genetic
algorithm is shown in Fig.4.1. This is compared
with the value obtained from actual implementation
i.e. obtained from hardware and simulation
implementation as shown in Table 4.2.
Fig.3.3 Validation Environment
IV.
RESULTS AND DISCUSSION
Here the graph of fitness function and number of
generations required is shown, whereas for
particular application what is fitness function value
is shown. Best value of fitness function is shown in
graph and for that fitness function what are the
parameter values that is also shown in MATLAB
optimization
toolbox.
Some
initialization
parameters are to be set in toolbox those are shown
in Table 4.1
The Fig.4.1 shows the graph where both parameters
values are taken on random basis and among that
best values are shown.
Using hardware
After
software co-
implementing GA
design
Computational
7.54 millisecond
9.599 millisecond
time / Iteration
time
Table 4.2 Comparison of GA obtained result and actual result
Here is comparison of area parameters and
computation time for one iteration whereas one
iteration is nothing but, time required to take one
turn and while taking turn how much time is
required to communicate with hardware(Spartan
3E500) kit. It is display on simulator as shown in
figure.
Sr.No.
Parameters
Values
1
Population size
20
2
Crossover rate
0.5
3
Mutation rate
0.2
4
Number of
51 for Fig.
Here Hardware opposition based PSO implemented
with ANN which utilizes more area as compared to
this implementation which is shown in Table 6.3(a)
whereas time required for software implementation
is also shown and compared for this application.
iterations
Table 4.1 GA parameter values
The lower limits and upper limits are shown in
previous chapter depending on that the values for
parameter x3 and for parameter x4 are calculated
which are 490.547 and 0.997 respectively.
ISSN: 2231-5381
http://www.ijettjournal.org
Page 81
International Journal of Engineering Trends and Technology (IJETT) – Volume 26 Number 2- August 2015
[4] Nadia Adnan Shiltagh, Lana Dalawr Jalal; Optimal Path
Planning For Intelligent Mobile Robot Navigation Using
Modified Particle Swarm Optimization. IJEAT 2013
Comparisons of area utilized by different
implementations
LUTs
Required
Hardware
oppositionbased PSO
applied to
mobile robot
controllers
[1]
Implementation of
soft computing
algorithm for
mobile robot
controller using
hardware software
co-design
43595 (63.1%)
597 (6%)
[5] Joon-Woo Lee, Jeong-Jung Kim, Byoung-Suk Choi, Ju-Jang
Lee; Improved Ant Colony Optimization Algorithm by Potential
Field Concept for Optimal Path Planning, IEEE-RAS
International Conference on Humanoid Robots, 2008
[6] Sudha, N, Mohan, A.R., Hardware-Efficient Image-Based
Robotic Path Planning in a Dynamic Environment and Its FPGA
Implementation. Industrial Electronics IEEE 2011
[7] Dr. R. Venkata Rao, Decision Making in Manufacturing
Environment Using Graph Theory and Fuzzy Multiple Attribute
Decision Making Methods, Springer 2012
[8] P. Raja* and S. Pugazhenthi, Optimal path planning of
mobile robots: A review, International Journal of Physical
Sciences Vol. 7(9), pp. 1314 – 1320, February 2012
20499(27.7%)
705(7%)
[9] Meshram, U. ; RKNEC, Nagpur, India ; Bande, P. ;
Dwaramwar, P.A. ; Harkare, R.R.Robot arm controller using
FPGA IEEE 2009
F/F
required
Time for
computation
1.01
millisecond
[10] Ms. Shilpa Kale, FPGA-based Controller for a Mobile
Robot (IJCSIS) International Journal of Computer Science and
Information Security, Vol. 3, No. 1, 2009
7.55 millisecond
[11] J. Gonzalez-Gomez, E. AguayoE. Boemo. Locomotion of a
Modular Worm-like Robot Using a FPGA-based Embedded
Micro Blaze Soft-processor. Springer 2005
Table 4.3 Comparison of area results with previous
implementations
V.
[12] Xuan Zou1, Bin Ge, Peng Sun, Improved Genetic
Algorithm for Dynamic Path planning. International Journal of
Information and Computer Science May 2012.
CONCLUSION
Hardware software co-design requires less
area and hardware resources than hardware
implementations with AI whereas takes less time
than other implementations as it is compared in
result tables. The comparison with other methods
shows that hardware software co-design is middle
way to overcome disadvantages of software and
hardware implementations separately in this
problem.
The parameter
optimization technique gives optimum values of
parameters which show better results for path
finding problem those values are shown in results
and it gives mathematical justification to obtained
results from hardware software co-design. Table
shows how computational time can be minimized
required for this application and at the same time
the area required also minimized which is shown in
Table.4.3
VI. REFERENCE
[1] Daniel M. Muñoz, Carlos H. Llanos B, Leandrodos S.
Coelho C, D ,Mauricio Ayala-Rincon E, F; Hardware
Opposition-Based PSO Applied To Mobile Robot Controllers.
Elsevier 2013
[2] M. Vijay, Dr. M. Jagadeeswari; An Efficient Architecture
For Robotic Path Planning. Ijarcsse 2012
[3] O. Hachour; the Proposed Genetic FPGA Implementation or
Path Planning of Autonomous Mobile Robot. International
journal of circuits, systems and signal processing 2008
ISSN: 2231-5381
http://www.ijettjournal.org
Page 82