Periodic robust robotic rock chop via virtual model control
This work addresses the challenge of contact-rich robotic cutting for manipulation tasks, offering a robust and platform-independent solution.
The paper introduces a virtual-model controller for robotic cutting that generates a robust, rhythmic rock-chop motion without pre-planned trajectories. Experiments with a Franka manipulator achieved sub-millimeter slice accuracy (1-6 mm thickness) at nearly one cut per second across five vegetables, with robustness to changes in knife shape, cutting board height, and different manipulators.
Robotic cutting is a challenging, contact-rich manipulation task where the robot must simultaneously negotiate unknown object mechanics, large contact forces, and precise motion requirements. Our hypothesis is that this complexity can be alleviated through the design of a physically structured virtual-model controller that uses switched virtual mechanisms to generate a robust, rhythmic rock-chop motion for robotic cutting, without requiring pre-planned trajectories or precise environmental information. Motion is generated by the interaction between the environment, the robot's dynamics, and the virtual forces of the switching virtual mechanism, ultimately realized through the available actuation. Through theoretical analysis and experimental validation, we demonstrate that the controlled robot behavior settles into a stable periodic motion. Experiments with a Franka manipulator demonstrate robust cuts across five different vegetables, achieving sub-millimeter slice accuracy for thicknesses from 1 mm to 6 mm at a rate of nearly one cut per second. The controller maintains high performance despite changes in knife shape or cutting board height, and successfully adapts to a different humanoid manipulator, demonstrating robustness and platform independence.