Authors: Amine Belaid; Boubekeur Mendil; Ali Djenadi
Addresses: Industrial Technologies and Information Laboratory, Faculty of Technology, Bejaia University, 06000 Bejaia, Algeria ' Industrial Technologies and Information Laboratory, Faculty of Technology, Bejaia University, 06000 Bejaia, Algeria ' Industrial Technologies and Information Laboratory, Faculty of Technology, Bejaia University, 06000 Bejaia, Algeria
Abstract: Rapidly exploring random tree star (RRT*) has been widely used for optimal path planning for the reason that can solve high degrees of freedom problems. However, this method has many limitations such as slow convergence rate and solving problems with narrow passages. In addition, the collision checking for this method consumes a lot of time in cluttered environments. In this paper, we present a new variant of RRT* named narrow passage RRT* (NP-RRT*), to deal mainly with narrow passage problems and cluttered environments. Our idea is to generate samples near obstacles to explore efficiently complex regions in the configuration space. We have also implemented a path optimisation technique to speed up the convergence rate. In order to reduce the complexity of collision checking, we used a pre-procedure that localises the obstacles before running the planning process. We demonstrate that the complexity of collision checking with our approach does not depend on a number of obstacles. Simulation results, performed in different environments comparing our algorithm with RRT*, alongside statistical analysis, confirm the efficiency of NP-RRT* method.
Keywords: path planning; collision checking; steer function; narrow passage; RRT*; sampling motion planning; optimal path; configuration space.
International Journal of Computational Vision and Robotics, 2022 Vol.12 No.1, pp.85 - 100
Received: 15 May 2020
Accepted: 26 Oct 2020
Published online: 30 Nov 2021 *