Genetic Trajectory Planner for a Manipulator with Acceleration Parametrization.

Abstract

This paper presents a genetic trajectory planning method of a robot manipulator producing the optimal trajectory between two end points. Genetic algorithm based methods seldom require a priori knowledge of a problem. Furthermore, they do not tend to fall into local optima and proceed toward the global optimum. However, they have difficulty in handling equality constraints of trajectory boundary conditions because they use probabilistic transition rules to find a solution. In this paper, we investigate the proper genetic trajectory parameterization and develop an efficient scheme for the implementation of genetic trajectory planner. We demonstrate the effectiveness and validity of the proposed approach through some simulation studies.