Trajectory Optimization Algorithm based on Robot Dynamics and Convex Optimization

Abstract

This paper mainly discusses the trajectory optimization problem of industrial robot along predetermined path in Cartesian space aiming at minimize the traverse time and smoothing the trajectory. The algorithm relies heavily on the complete robot dynamics model, in which both viscous friction and Coulomb friction are included. Based on convex optimization theory, the trajectory optimization formulation can transform into a second-order cone programming (SOCP) problem, which can be solved numerically through a discretization scheme and SDPT3 solver. Additionally, a constraint generated from the torque-speed curve of motors is added to the aforementioned convex optimization formulation. The simulation results show that our improved trajectory optimization algorithm attains smooth and time optimal trajectory with shorter traverse time and good tracking performance compared with previous methods.