Improved Accuracy and Contact Stability in Robotic Contouring With Simultaneous Registration and Machining

Poor workpiece registration is a limiting factor in robotic machining. Force control can correct for path errors; however, controller tuning is difficult as machining quality depends on disparate goals. Fast edge-tracking requires low damping, while maintaining stable tool contact requires high damping. We introduce Simultaneous Registration and Machining (SRAM), a novel framework to improve robotic machining performance in contouring applications. SRAM uses force and position feedback during machining to improve its registration estimate and apply real-time path corrections. Simultaneously, controller damping is modulated based on the registration covariance. Thus, the controller rapidly corrects for tracking error when registration is uncertain, but transitions to stable behavior when possible for optimal finish quality. The algorithm is validated in robotic deburring testing, showing an 88% reduction in path error and virtually eliminating force-tracking errors compared with a nominal controller. Machining quality is improved and tool wear notably decreased. SRAM lowers the required registration accuracy while improving machining quality, reducing cost and cycle times.