Force Feedback and Tactile Sensing for Robotic Teleoperation of Contact Rich Manipulation Tasks

Abstract

Imitation learning has shown promising results in teaching robots new skills. We propose augmenting the ALOHA bimanual teleoperation system with haptic feedback to obtain higher quality expert demonstrations. We add two types of haptic feedback: force feedback and cutaneous feedback in both a real and simulation teleoperation system. Additionally, we propose to add tactile sensors to observe the impact of tactile data to imitation learning models in solving fine manipulation tasks.