Joint Torque Estimation using Base Force-Torque Sensor to Facilitate Physical Human-Robot Interaction (pHRI)

• Main Authors: Das, Sumit Kumar, Saadatzi, Mohammad Nasser, Abubakar, Shamsudeen, Popa, Dan O.
• Format: Conference Proceeding
• Language: English
• Subjects: Artificial neural networks; Collision avoidance; Gravity; Manipulator dynamics; Robot sensing systems
• ISSN: 2161-8089
• DOI: 10.1109/COASE.2019.8843092

To detect forces during physical Human-Robot Interaction (pHRI), a force-torque sensor (FTS) is generally attached at the wrist of a robot manipulator. Alternatively, collaborative robots can measure interaction forces via torque sensing at their joints. Yet another direction toward safe and interactive robots is to cover them in smart skins with embedded tactile sensors. In this paper, we explore another idea to facilitate pHRI using an FTS placed at the base of a robot arm. The resulting base force-torque sensor (BFTS) is able to sense external forces and torques applied anywhere along the robot body. We formulate a model-free, on-line learning controller to estimate the interaction forces on the robot from the BFTS data. The controller does not require a robot dynamic model to operate, and has Lyapunov stability guarantees. We conduct experiments to validate the mean-square estimation error of our scheme using a custom 6-DOF robotic arm under real-time control. Results show that the measured torques at individual joints closely follow the estimated values. In the future, this controller can be used for adaptive pHRI with non-collaborative robots or robot manipulators.