Methodological approach for workspace boundary determination in joint space using support vector machine

Abstract

For most industrial robots using position control, self-collision is a headache. Without well-defined constraints at a joint level, self-collision can lead to damage to a robot or even injury to the operator. In this study, a new methodological approach is proposed that is different from the traditional methods of using sensors or analysing kinematics. The proposed method uses a support vector machine to construct a joint boundary map based on the robot's movement in the workspace. To verify its feasibility and performance, simulations are conducted. The proposed method is believed to help reduce the cost and computational burden. It is expected to be utilised even in applications where the existing methods are difficult to use.