Page 258 - 2024-Vol20-Issue2
P. 258
254 | Sabeeh & Al-Furati
path-planning. A more efficient algorithm can reduce TABLE I. PERFORMANCE COMPARISON BETWEEN
planning delays, allowing the robot to quickly respond GA-PRM ALGORITHM AND A*, RRT, GENETIC, AND
to changes in the environment and dynamic obstacles.
PRM ALGORITHMS
• Average smoothness measures: the continuity and
absence of abrupt changes in the robot’s motion during No. Algorithm APL ACT AS
navigation. Paths with smoother trajectories lead to
more stable and comfortable robot movements. This 1 Proposed GA-PRM 25.6235 0.6881 0.3133
is especially important when the robot is interacting
with humans or in the presence of dynamic obstacles. A 2 A* 29.1758 0.7452 0.0803
smoother path reduces the risk of collisions, improves
user comfort, and ensures safer robot navigation in 3 RRT 36.2037 0.6209 0.2911
complex environments.
4 Genetic 37.43 0.7147 1.5308
VI. RESULTS
5 PRM 26.8700 0.9962 1.8543
A. Result of Simulation
Fig. 5 visually represents the robot’s movement in the simu- naturally handle dynamic obstacles or provide probabilistic
lated workspace. The workspace is divided into four sections, roadmaps for path planning [45]. Table I provides a com-
each corresponding to a different quadrant. Subfigure (2-A) parison between the GA-PRM algorithm and four other path-
shows the robot’s position in quadrant 1, (2-B) in quadrant planning algorithms (A*, RRT, Genetic algorithm, and PRM).
2, (2-C) in quadrant 3, and (2-D) in quadrant 4. Each part of This comparison is based on three performance metrics: av-
the diagram illustrates the robot’s path as it moves through its erage path length, average computational time, and average
respective workspace section. Gray boxes represent stationary smoothness.
obstacles, while small black dots depict moving obstacles.
As can be seen from Table I, the GA-PRM algorithm
B. Result of Comparison achieves the shortest average path length among all the al-
Fifty tests were conducted to compare the performance of gorithms, with an APL of 25.6235 units. This indicates that
four path-planning algorithms: A*, RRT, Genetic, and PRM. the current algorithm is successful in finding paths that are,
In each test, the robot had to move from the starting point on average, shorter than those generated by the other algo-
to the target point in a predefined area. The same specific rithms. While GA-PRM excels in shortest average path length,
parameters, workspace size, grid size, fixed and moving ob- it requires a moderate amount of computational time, with an
stacles, and sensor settings were used for all four algorithms. ACT of 0.6881 seconds. This indicates that it strikes a balance
The choice of comparing the GA-PRM algorithm with the A*, between path length and computational efficiency. In addi-
RRT, Genetic, and PRM algorithms is based on their well- tion, GA-PRM produces paths with a relatively high average
established effectiveness and relevance in motion plansning smoothness (AS of 0.3133). This suggests that it achieves a
and optimization. A* stands out for its efficiency in finding the good balance between path length and smoothness, resulting
shortest paths in grid-based environments [43]. RRT excels in paths with less abrupt changes in direction.
in high-dimensional spaces with dynamic obstacles due to its
probabilistic completeness and fast convergence [1]. Genetic The GA-PRM algorithm demonstrates notable strengths
algorithms offer versatility in optimizing complex spaces, that make it well suited for deployment in a hospital environ-
making them valuable for benchmarking global optimization ment. Firstly, it excels in generating shorter and smoother
by the GA-PRM algorithm [44]. PRM, a sampling-based paths, which can be pivotal in healthcare settings where preci-
method, is known for its simplicity and efficiency in roadmap sion and patient safety are paramount. These characteristics
construction, making it a suitable comparison for evaluating contribute to minimizing the time taken for robots to navigate
the GA-PRM algorithm’s roadmap generation performance through hospital corridors and reduce the risk of unexpected
in high-dimensional spaces [3]. Other algorithms, while fun- obstacles. Secondly, the algorithm’s ability to optimize path
damental and well-established path-finding algorithms, may smoothness ensures that robotic movements are fluid and less
have characteristics that make them less suitable for direct likely to cause disruptions or discomfort to patients, staff, and
comparison in the context of my research. For example, Di- visitors. Although it may exhibit slightly longer computa-
jkstra’s Algorithm is known for its optimality in finding the tion times, the trade-off is justifiable in healthcare, given the
shortest path in static environments. However, it does not emphasis on safe and efficient navigation within a controlled
and predictable environment. Overall, the GA-PRM algo-
rithm aligns well with the requirements of a hospital setting
by prioritizing path quality and patient well-being.
In Fig. 6, a comparison is made among various path-
planning algorithms using three key performance measures
given above. The algorithms under evaluation include GA-
PRM, A*, RRT, Genetic, and PRM. The chart illustrates these
