•  
  •  
 

Corresponding Author

Mohammed T. Hameed

Document Type

Article

Keywords

Path planning, Lynxmoition AL-5D, Rapidly-exploring random tree, Low-discrepancy sequences, Free Cartesian Space (FCS)

Abstract

Path planning is a fundamental task in robotic systems, especially for manipulators operating in constrained environments. The use of low-discrepancy sequence techniques in traditional planning algorithms can significantly improve performance, particularly in terms of speed of convergence and optimal path. In this work, we utilize the popular quasi-random sequence methods, the Halton and Sobol sequences, to explore their impact on the Rapidly Exploring Random Tree (RRT) algorithm. The APF-RT*-HS and the APF-RT*-SB are path-planning algorithms that are experimentally compared in this paper in terms of path length and the number of iterations, which are essential factors affecting robot energy efficiency and responsiveness. A modified version of the Lynxmotion AL5D robot arm was used as a case study in three different static test environments, representing complex environments frequently encountered in industrial automation work. Based on the experimental test results, The APF-IRRT*-SB algorithm achieved improvements of approximately 3.04% in path length and 9.14% in the number of iterations compared with the APF-IRRT*-HS. This improvement translates into shorter path lengths and reduced computation time, which are crucial for enabling robotic arms to operate more efficiently in cluttered industrial environments and real-time applications. These results indicate that using the Sobol sequence in path planning for robots produces more efficient paths compared to using the Halton sequence.

Download

Share

COinS