Learning force-based manipulation of deformable objects from multiple demonstrations

Manipulation of deformable objects often requires a robot to apply specific forces to bring the object into the desired configuration. For instance, tightening a knot requires pulling on the ends, flattening an article of clothing requires smoothing out wrinkles, and erasing a whiteboard requires applying downward pressure. We present a method for learning force-based manipulation skills from demonstrations. Our approach uses non-rigid registration to compute a warping function that transforms both the end-effector poses and forces in each demonstration into the current scene, based on the configuration of the object. Our method then uses the variation between the demonstrations to extract a single trajectory, along with time-varying feedback gains that determine how much to match poses or forces. This results in a learned variable-impedance control strategy that trades off force and position errors, providing for the right level of compliance that applies the necessary forces at each stage of the motion. We evaluate our approach by tying knots in rope, flattening towels, and erasing a whiteboard.

Authors

• Alex X. Lee
• Henry Lu
• Abhishek Gupta 0004
• Sergey Levine
• Pieter Abbeel

Keywords

• Trajectory
• Force
• Robots
• Three-dimensional displays
• Joints
• Impedance
• Kinematics
• Deformable Objects
• Wrinkles
• Position Error
• Feedback Gain
• Non-rigid Registration
• Motion Stage
• Force Error
• Configuration Of Objects
• Warping Function
• Time Step
• Point Cloud
• Joint Angles
• Gaussian Mixture Model
• Behavior Use
• Joint Space
• Reward Function
• Force Sensor
• Cm In Size
• Maximum A Posteriori
• Inverse Reinforcement Learning
• Stage Of The Task
• Force Variability
• Dynamic Time Warping
• Thin-plate Spline
• Point Cloud Registration
• Force Term
• Empirical Covariance
• Time Warping
• Time And Space