Trajectory Planning of a Six-DOF Robot Based on a Hybrid Optimization Algorithm

This work adopts a standard Denavit-Hartenberg method to model a PUMA 560 spot welding robot as the object of study. The forward and inverse kinematics solutions are then analyzed. To address the shortcomings of the ant colony algorithm, factors from the particle swarm optimization and the genetic algorithm are introduced into this algorithm. Subsequently, the resulting hybrid algorithm and the ant colony algorithm are used to conduct trajectory planning in the shortest path. Experimental data and simulation result show that the hybrid algorithm is significantly better in terms of initial solution speed and optimal solution quality than the ant colony algorithm. The feasibility and effectiveness of the hybrid algorithm in the trajectory planning of a robot are thus verified.

I. Introduction

Robot trajectory planning, which is one of the basic processes involved in the operation of robots, affects the stability and accuracy of a robot. It is an indispensable aspect of a robot control system, and has continuously gained considerable attention from researchers. Robot trajectory planning comprises Cartesian space planning and joint space planning. The latter has the advantage of easily fulfilling design requirements and motion constraints because the control system directly affects the joint, thereby avoiding singular point and joint mutation. In addition, the computation of the inverse kinematics of a redundant manipulator is unnecessary. Nevertheless, joint space planning is used only for jobs without path requirements. Cartesian space planning is visual and intuitive. However, fast inverse kinematics calculation should be realized within robot control. Moreover, this approach cannot ensure the lack of singularity. For jobs with strict requirements on the instantaneous variation law of path and pose, such as continuous spraying operations, trajectory planning must be performed in Cartesian coordinate space [1].