Design of jerk bounded trajectories for online industrial robot applications

Abstract

An online method for obtaining smooth, jerk-bounded trajectories has been developed and implemented. Jerk limitation is important in industrial robot applications, since it results in improved path tracking and reduced wear on the robot. The method described herein uses a concatenation of fifth-order polynomials to provide a smooth trajectory between two points. The trajectory is determined based on approximating a linear segment with parabolic blends trajectory. A sine wave approximation is used to ramp from zero acceleration to non-zero acceleration. This results in a controlled quintic trajectory which does not oscillate, and is near time-optimal given the jerk and acceleration limits specified. The method requires only the computation of the quintic control points, up to a maximum of seven points per trajectory way-point. This provides hard bounds for online motion algorithm computation time. Simulations and experimental results on an industrial robot are presented.