Real-Time Model Predictive Control for Energy-Optimal Obstacle Avoidance in Parallel SCARA Robot for a Pick and Place Application

摘要

This article presents a nonlinear model predictive control (NL-MPC) design and real-time implementation for energy-optimal obstacle avoidance of a parallel Selective Compliance Articulated Robot Arm (SCARA) robot for a pick and place application. The NL-MPC problem is solved online using a real-time iteration scheme that guarantees the computational time within the solving time for the pick and place application with obstacle avoidance. The proposed approach is tested on a conveyor belt simulation scenario with dynamic obstacles and compared with a conventional industrial application solution. The results show that the NL-MPC approach can improve the energy efficiency by approx. 25% while avoiding the obstacles successfully.