Discrete dynamic programming for optimized path planning of flexible robots

Abstract

An optimized path-planning approach is presented for flexible dynamic systems. Feedforward command profiles are determined for rest-to-rest large angle maneuvers that keep vibrations to a minimum. The performance index included both minimum energy and minimum time optimization problems. The feedfoward command profiles were generated by solving a discrete-time optimal control problem via discrete dynamic programming (DDP). A simple planar two-link flexible robot arm was utilized to demonstrate the DDP path-planning scheme. Numerical simulation results included: 1) minimum effort, 2) minimum effort with bounds, 3) minimum time, and 4) minimum torque-rate or power, respectively. The DDP approach demonstrated an effective alternative for numerical optimization, to generate optimal feedforward trajectory profiles.