Chapter 5: Conclusion
Chapter 5: Conclusion
Chapter 5: Conclusion
defence, mining, hospitals, etc. Efficient path planning algorithms are required for
situations even though the optimization criteria may be the same viz. path length or travel
time, what is critical is the computation time required to plan a collision-free path
segment. Therefore, this work fulfills the twin objectives of developing effective off-line
path planning algorithm which gives the shortest path and developing computationally
For a given known environment, in order that the mobile robot reaches its target, an
initial population of valid alternative paths is generated. A path is valid if it does not
interfere with any obstacle (static or moving). In the process of generating the required
size of population, an invalid path interfering with a stationary obstacle is converted into
valid one by gliding along the edges of the amplified obstacle. In case of an invalid path
valid path as it will cause undue delay in initial population generation itself.
82
The valid paths are then subjected to evaluation function to calculate its fitness. PSO
technique is used by the algorithm to select collision-free optimal paths. The evaluation
function is designed to have shortest path length as an objective. The proposed algorithm
quality of initial population. The effectiveness of the proposed algorithm has been shown
obtained with GA in place of PSO. The obstacles in the environment are designed to
contain convex, concave and curved obstacles. Results show that the algorithm using
PSO delivers marginally shorter path lengths. More importantly, in terms of computation
time over 29% improvement has been achieved when compared to GA.
which are sparsely located or cluttered. Comparative performance of the algorithm shows
that a minimum of 1.3% reduction in path length and a maximum of 8.9% reduction in
path length are achieved for sparse and cluttered dynamic environments respectively.
More importantly, the computation time required by the proposed algorithm is only
around 1% of the time required by Wang et al. algorithm. Also, the path obtained by the
proposed algorithm has less number of path segments. The comparison of results show
that the proposed algorithm is very effective for cluttered dynamic environments.
83
Also, the ruggedness of the algorithm is demonstrated using Monte Carlo simulation
trials. Five different tasks are designed for each of the four simulated environments to
obstacles. After 1000 trials each in the 20 different situations, only around 1% deviation
is found in optimal path length. Thus the ruggedness of the algorithm is proved.
The salient features of the proposed off-line path planning algorithm which yields better
1. An open initial search space without limiting to the vertices of the enlarged
2. Generation of valid paths using direction concept without using any special
at the minimum and n 2 at the maximum where n is the total number of vertices
of all the obstacles in the environment. Thus, the path is modelled according to
4. Particle Swarm Optimization technique with real strings is used by the algorithm
for choosing the shortest path. Real coded formulation of the problem is much
84
easier than the binary representation. This eliminates the need for coding and
An effective on-line path planning algorithm for mobile robots in dynamic environments
considers all the current on-line information of robot as well as nearing obstacles. The
kinematic and dynamic constraints of robot. The effect of constraining obstacles is also
taken into consideration while negotiating the most imminent obstacle. The proposed
algorithm does not require any separate recovery mode approach to escape from trap
considered in chapter 3, here also the environments have been designed to be sparse in
two cases and cluttered in the remaining two cases containing obstacles of different sizes.
In addition, the performance of the algorithm is compared with Min et al. algorithm for
performance of the algorithm shows that the proposed algorithm is computationally more
efficient and effective in reaching targets along shortest possible paths. Around 30%
85
reduction in path length and 70% reduction in computation time are achieved for
Also, the ruggedness of the algorithm has been demonstrated using Monte Carlo
simulation trials. Five different tasks are considered for each sparse and cluttered
respective targets. Results such as computation time per instant of the robot and its
standard deviation are presented for 1000 simulation trials for all the 20 cases. In terms of
path length as well as computation time, less than 1% deviation has been achieved in all
the cases.
The salient features of the proposed on-line path planning algorithm to yield shorter paths
obstacles.
2. The influence of other constraining obstacles while negotiating the most imminent
obstacle is also included in the model to take care of further deviation required
3. Kinematic and dynamic constraints of robot are also built in the mathematical
model.
86
4. The need for trap recovery is eliminated as the path is planned tangential to the
collision zone.
algorithms have been developed for mobile robots in dynamic environments. However,
there is scope for further improvements. For off-line path planning, the objective may
have multiple optimization criteria with varying weightages for improving quality of
paths. For on-line path planning, this work can further be extended to applications
involving multiple robots moving towards different targets or same targets, which can
also be moving.