Automatic Synthesis of a Saturating State Space Controller Based on Convex Optimization for Industrial Robots

Abstract:
Robotic control needs to be optimal if it is to stay competitive in industrial production scenarios. However, highly tweaked controllers cannot directly transfer from one robot setup to another. Thus, the controllers must be automatically generated concerning their use in specific production processes. This paper presents an automatic synthesis of an almost time-optimal model-based saturating state space controller. The control parameters are derived from the solution of a convex optimization problem. The approach complements previous work of transferable skill-based programming for robotic assembly in particular. This paper introduces the problem formulation, exercises the optimization on a task space controlled industrial robot, and finally evaluates it in simulation and compares experimental results with a state-of-the-art industrial robot controller.