Improved A* Algorithms For DDMR With Static And Dynamic Constraints With Comparison Study Of DWA, Dijkstra, RRT, And Traditional A*

Abstract

This paper presents a path planning method based on the traditional A* algorithm for a two-wheeled Non-Holonomic Differential Drive Mobile Robot (DDMR). The proposed method supports autonomous navigation and obstacle avoidance and also diminishes the drawbacks of A* algorithm, such as large turning angles in the robot's path, unsmooth trajectories, and applicability only to static environments. This method adopts A* algorithm and a weighted heuristic function incorporating curvature. The weighted heuristic function decreases the path length, while the curvature function smooths the path. Three different scenarios (maps) is simulated to evaluate the proposed method. Robot Operating System (ROS), Gazebo and RViz is used to simulate three different scenarios with consistent navigation parameters for the mentioned methods. The results obtained were compared with those of DWA, Dijkstra, RRT, and A* algorithm. The comparison proved the superiority of the proposed method, in removing redundant tipping points to smooth the planned path and shortening its length 60%. The effect of weights and curvature angles on the distance function, as well as the effects of local and global cost map parameters on the proposed method is studied and clarified.