Motion Planning Based on Genetic Algorithm for Manipulator Provided in ROVs

Abstract

This paper describes the design method used in motion planning for a manipulator provided in a Remotely Operated Vehicle (ROV) . The method incorporates path planning for the manipulator tip and posture planning for the manipulator in a workspace divided into cubes. Each type of planning employs two kinds of genetic algorithms. One is used to generate suitable tip positions. The other is used to generate candidates for a secure manipulator posture that will be consistent with the tip positions. The secure manipulator posture, which avoids obstacles, is chosen from among generated candidates on the basis of an evaluation considering the drag force. The manipulator motion is planned by repeatedly executing this process. To confirm the validity of this method, an experiment was performed using the manipulator with seven degrees of freedom in an environment that included large obstacles.