Multi-Feature Enhancement of Standard and Modified A-star and RRT Algorithms

Abstract

This manuscript suggests improvements in offline path planning algorithms: A-star (A*) and Rapidly Exploring Random Trees (RRT)- well-celebrated algorithms for robot path planning. The conventional algorithm of A* is reliable for path length efficiency, while RRT focuses on relaxing computation complexity. Specifically, this work aims to introduce three feature enhancements: bi-directionality, obstacle padding, and smoothing—to improve the capability of traditional algorithms. These enhancements have been applied to A*, a modified version of RRT, and the jump-point search (JPS) variant of A*. A comparative study is presented for the path-planning of a nonholonomic autonomous mobile robot to assess the effectiveness of these modified algorithms with augmented features. The simulations were conducted using MATLAB, where a configuration space with multiple obstacles is considered. Finally, insights into the advantages offered by these enhancements and their impact on the performance of the algorithms have been discussed in detail