An Efficient Iterative Learning Approach to Time-Optimal Path Tracking for Industrial Robots

摘要

In pursuit of the time-optimal motion of an industrial robot along a desired path, a previously identified model is typically used to calculate the required inputs for perfect tracking. An inevitable model-plant mismatch, however, causes the obtained inputs to be suboptimal-resulting in poor tracking performance-or even be infeasible by exceeding given limits. This paper, at hand, presents a two-step iterative learning algorithm that compensates for such model-plant mismatch and finds the time-optimal motion, improving tracking performance, and ensuring feasibility. Due to an efficient solution of the path tracking problem using a sequential convex log barrier method, the delay between consecutive task executions is eliminated. To show the effectiveness of the proposed algorithm, an experimental validation on a standard industrial manipulator is performed, illustrating that the developed approach is capable of reducing the execution time while at the same time improving the tracking performance.