Page 260 - 2024-Vol20-Issue2
P. 260
256 | Sabeeh & Al-Furati
Fig. 6. Path-planning with GA-PRM algorithm. ing performance in experiments, there are still areas for im-
provement.
more direct and efficient in reaching the goal while
avoiding obstacles. The GA-PRM’s combination of ge- First, the algorithm’s performance depends heavily on the
netic algorithms and probabilistic roadmaps contributed choice of parameters, such as the mutation rate and population
to the algorithm’s ability to explore the workspace ef- size. These parameters can significantly influence the quality
fectively and find shorter paths. This characteristic is of of the generated paths. Fine-tuning the genetic algorithm’s
utmost importance in various autonomous robotic ap- parameters and exploring alternative genetic operators could
plications, as shorter paths translate to reduced energy further improve the algorithm’s convergence and solution
consumption and faster completion of tasks, which can quality.
be crucial for time-sensitive missions.
Second, as the environment becomes more complex and
2. Computational time: The GA-PRM algorithm demon- the count of dynamic obstacles rises, the algorithm’s execu-
strated an average computational time of (0.6881), which tion time also experiences an increase. Although genetic algo-
was faster in generating path plans as compared to the rithms and probabilistic roadmaps naturally bring in elements
other algorithms (except for the RRT algorithm, which of randomness and adaptability, it remains crucial to inves-
took 0.6209 seconds). The efficiency of the GA-PRM tigate strategies that can enhance computational efficiency
algorithm can be attributed to its use of genetic algo- when dealing with larger and dynamic settings.
rithms and probabilistic roadmaps, which allowed for
effective exploration of the configuration space while Another aspect to consider is how to handle dynamic
keeping the computation time low. This characteristic obstacles with unpredictable trajectories. The current algo-
makes the new algorithm suitable for real-time applica- rithm models dynamic obstacles as random walkers within
tions where prompt decision-making is imperative. the workspace. However, incorporating predictive methods
or learning algorithms to anticipate the future trajectories of
3. Smoothness: The GA-PRM algorithm achieved a mod- these obstacles could lead to more predictive and preemptive
erate level of smoothness (0.3133) in its planned paths. path-planning behavior for the robot.
While the PRM algorithm had the smoothest paths with
the highest average smoothness (1.8543), the GA-PRM Finally, the proposed algorithm currently assumes known
algorithm managed to strike a balance between path and fixed sensor ranges for ultrasonic, IR, and other sensor
smoothness and path length, leading to an overall more types. Incorporating adaptive sensor models that can dynami-
optimal solution. Smoothness in the path is crucial to cally adjust their ranges based on the environment’s charac-
guaranteeing stable and controlled movement of the teristics could enhance the algorithm’s robustness in handling
robot, particularly in situations with strict safety de- varying obstacle densities.
mands.
VIII. CONCLUSION
Although the proposed GA-PRM algorithm has shown promis-
The GA-PRM algorithm holds significant importance in robotics
and path planning, particularly in dynamic and complex en-
vironments such as healthcare settings. Through combin-
ing the power of genetic algorithms with the efficiency of
probabilistic roadmaps, GA-PRM excels in finding adaptable
and obstacle-aware paths for robots. This characteristic is
crucial for ensuring safe and efficient navigation in environ-
ments where obstacles and conditions are subject to frequent
changes, ultimately contributing to the reliable and effective
deployment of robots in real-world scenarios, including health-
care, logistics, and more.
Future research could optimize GA-PRM parameters, in-
tegrate advanced sensors for context-aware planning, explore
multi-robot systems, and leverage hardware advancements for
real-time implementation.
CONFLICT OF INTEREST
The authors have no conflict of relevant interest to this article.
