Car‐like robot motion planning based on grey relational pattern analysis

Abstract

This paper proposes a simple exploring approach for a car‐like robot to determine a near‐optimal path between prescribed initial and goal positions, based on the nonlinear programming problem and grey relational pattern analysis. No matter whether the considered workspace is known or not in advance, the proposed approach can make the car‐like robot explore and move in a workspace containing multiple convex obstacles. Unlike other exploring approaches, the proposed find‐path procedure must consist of at least one trial. Each trial contains two main stages, one is termed the forward search stage and the other is named the backward search stage. After all possible trials have occurred, an additional stage, called the decision‐making stage, is then introduced to determine the near‐optimal and collision‐free path, which is guaranteed to reach the goal. In addition, the presented method is applicable for on‐line planning applications and, furthermore, can solve the so‐called local minimum problems. Simulation results for a workspace with multiple convex obstacles demonstrate the searching performance of our approach and its potential as an on‐line path planner in a known or an unknown workspace.

Key words:

• car‐like robot motion planning
• grey relational pattern analysis
• nonlinear programming problem
• convex obstacle

Notes

Corresponding author. (Tel: [phone] ext. 3470 re‐ext 801; Fax: [phone] ext 3470; Email: [email protected])