AI summaryⓘ
The authors tackle the problem of controlling robots that need to touch and work on objects carefully, like in industrial deburring where a tool must be inserted precisely and moved smoothly while avoiding collisions. They combine two approaches: one that predicts safe robot motions using learned movement patterns (diffusion motion priors) and another that actively adjusts the robot’s force and movements in real time (force-feedback MPC). Their method helps robots perform delicate tasks accurately even in tight or tricky spaces. They tested their system on a robot and showed it could maintain the right force and avoid obstacles effectively. This is the first work to combine these techniques for such complex, contact-rich tasks.
Model Predictive Control (MPC)Force FeedbackDeburringTorque-Controlled RobotsDiffusion ModelsMotion PriorsCollision AvoidanceContact-Rich TasksTool InsertionIndustrial Robotics
Authors
Krzysztof Wojciechowski, Ege Gursoy, Arthur Haffemayer, Sebastien Kleff, Vincent Bonnet, Florent Lamiraux, Nicolas Mansard
Abstract
Model Predictive Control (MPC) is widely used for torque-controlled robots, but classical formulations often neglect real-time force feedback and struggle with contact-rich industrial tasks under collision constraints. Deburring in particular requires precise tool insertion, stable force regulation, and collision-free circular motions in challenging configurations, which exceeds the capability of standard MPC pipelines. We propose a framework that integrates force-feedback MPC with diffusion-based motion priors to address these challenges. The diffusion model serves as a memory of motion strategies, providing robust initialization and adaptation across multiple task instances, while MPC ensures safe execution with explicit force tracking, torque feasibility, and collision avoidance. We validate our approach on a torque-controlled manipulator performing industrial deburring tasks. Experiments demonstrate reliable tool insertion, accurate normal force tracking, and circular deburring motions even in hard-to-reach configurations and under obstacle constraints. To our knowledge, this is the first integration of diffusion motion priors with force-feedback MPC for collision-aware, contact-rich industrial tasks.