Multi-objective optimal trajectory planning of customized industrial robot based on reliable dynamic identification for improving control accuracy

Abstract

Purpose The purpose of this paper is to enhance control accuracy, energy efficiency and productivity of customized industrial robots by the proposed multi-objective trajectory optimization approach. To obtain accurate dynamic matching torques of the robot joints with optimal motion, an improved dynamic model built by a novel parameter identification method has been proposed. Design/methodology/approach This paper proposes a novel multi-objective optimal approach to minimize the time and energy consumption of robot trajectory. First, the authors develop a reliable dynamic parameters identification method to obtain joint torques for formulating the normalized energy optimization function and dynamic constraints. Then, optimal trajectory variables are solved by converting the objective function into relaxation constraints based on second-order cone programming and Runge–Kutta discrete method to reduce the solving complexity. Findings Extensive experiments via simulation and in real customized robots are conducted. The results of this paper illustrate that the accuracy of joint torque predicted by the proposed model increases by 28.79% to 79.05% over the simplified models used in existing optimization studies. Meanwhile, under the same solving efficiency, the proposed optimization trajectory consumes a shorter time and less energy compared with the existing optimization ones and the polynomial trajectory. Originality/value A novel time-energy consumption optimal trajectory planning method based on dynamic identification is proposed. Most existing optimization methods neglect the effect of dynamic model reliability on energy efficiency optimization. A novel parameter identification approach and a complete dynamic torque model are proposed. Experimental results of dynamic matching torques verify that the control accuracy of optimal robot motion can be significantly improved by the proposed model.

Multi-Objective Optimal Trajectory Planning for Robotic Arms Using Deep Reinforcement Learning

“…Liu [16] introduced a trajectory competition multiobjective particle swarm optimization algorithm, addressing cooperative robot trajectory planning from the three aspects of time, energy, and pulse optimization. Hou [17] proposed an innovative time-energy consumption optimization trajectory-planning method for dynamic identification. Rout [18] employed a multi-objective ant lion optimization technique to obtain the optimal trajectory by minimizing the time-torque ratio.…”