Page 101 - 2023-Vol19-Issue2

97 | Al-Ansarry, Al-Darraji & Honi

TABLE I.

PLANNING METHODS COMPARISON (STATIC MAP)
Method Time (sec.) Cost (node) Length

A* 3.40 311 350 cm

Dijkstra 7.88 453 365 cm

Dijkstra-PF 6.04 440 335 cm

respectively. Figure (7) shows the static map with global
paths that generates using the aforementioned methods above.

B. Experiment-2: (Local Path - Dynamic)
In this experiment, the proposed (Dynamic D-PF) has been
tested on a dynamic map of size (600*600 cm) and the results
of (50 runs) have been evaluated. As demonstrated in Table II,
the range of both time and cost values was reasonable between
the highest and lowest value. Finally, the standard deviation
was low and the median value average was also convincing.
In this experiment, the Dynamic D-PF has the ability to deal
with the dynamic obstacles based on employing the potential
field in case of facing moving obstacles accidentally where
the forces of repulsive and attractive guide the robot away
from the obstacles and move toward the goal by selecting
the nearest alternative safe path to be adopted. Figure (8)
simulates the dynamic path planning.

TABLE II.

DYNAMIC PLANNING BASED ON ROBOT VELOCITY (8

CM/SEC) AND THE DECISION TIME
Dynamic D-PF Time (sec.) Cost (node) Length

STD. 2.2 34.6 18.3 cm

Mean 114.5 1040 820 cm

Median 113.2 1025 810 cm

Lowest Value 112 1004 800 cm

The results of the experiments showed that the proposed Fig. 7. Offline Path Planning (a) Static Map (b) A* Path (c)
method (Dynamic D-PF) performed well in terms of path Dijkstra Path (d) Dijkstra-PF Path
length, safety, execution time, and demonstrating its effective-
ness as a path planning solution in dynamic 2D environments. (spaces) within the given environment. Although the A* algo-
In comparison, the proposed method was found to provide a rithm utilizes a heuristic function to identify a single optimal
good balance between (A* and Dijkstra) based on path length, path without the need to explore the entire map, it requires
safety, and execution time. Moreover, the Dynamic D-PF in less time than the Dijkstra algorithm. However, in the context
this stage can deal with the dynamic obstacles (moving obsta- of this work, it is necessary to create and store the desired
cles that appear accidentally) due to the benefit of merging the paths, except for those that extend far from the goal. This
Potential Field (PF) and Dijkstra which makes the Dynamic is done to provide alternative paths that can be used when
D-PF detects the obstacles efficiently and find another safe dynamic obstacles unexpectedly intersect with the global path
path quickly.

The algorithm of Dijkstra is capable of generating the short-
est optimal path by searching all the available configurations