Performance analysis of A*, Dijkstra and RRT path planning algorithms on ROS-based Gazebo for LiDAR-based mecanum wheeled mobile autonomous robot

Effective path planning is crucial for mobile robots to navigate safely and efficiently in various environments. The main purpose of this study is to compare and investigate the performances of three important path planning algorithms, namely A*, Dijkstra and Rapidly exploring Random Trees (RRT), for a mobile autonomous robot with LiDAR sensors and mechanical wheels. The mobile robot can move in multiple directions with its mecanum wheels and can precisely avoid obstacles with the help of LiDAR sensors and decision-making mechanisms (path planning algorithms). Various scenarios with cluttered and open areas were used in a simulated ROS Gazebo environment to evaluate the effectiveness of each algorithm. The performance metrics among the algorithms were analyzed with respect to path length, time to reach the target, velocity command frequency and CPU utilization. In terms of travel time performance criterion, A* performed approximately 35% better than Dijkstra and 85% better than RRT. In terms of the path length traveled, A* reached the target in approximately 11% shorter path length than Dijkstra and 17% shorter path length than RRT. In terms of the number of velocity commands processed, A* outperformed Dijkstra by approximately 36% and RRT by 38%. In terms of CPU utilization performance criterion, RRT performed approximately 10% better than A* and 74% better than Dijkstra. As a result, significant information was obtained about the strengths and weaknesses of each algorithm in selecting the most appropriate path planning strategies for mobile autonomous robots.