Plenary Talk 06 - Clutching Magnetorheological Actuators: An Enabling Technology for Haptic- Robots and High-Performance Machines
My Session Status
Speaker: Prof. Jean-Sébastien Plante, University of Sherbrook, Canada
Chair: Dr. Andreas M. Menzel
Abstract: In today's era of artificial intelligence, the progress of human-robot physical interaction is now limited by robotic hardware, notably, actuators. Unlike human muscles, the gearmotors used in most robots on the planet cannot perform strong, powerful tasks as well as gentle, dexterous ones in the same package. This presentation will show how clutching actuators based on magnetorheological (MR) fluid clutches repel conventional gearmotor limitations and open the door to new "haptic-robots" capabilities. Functioning of the MR clutching technology is first explained and current applications to active suspension seats and automotive active suspensions are reviewed. Then, an analytical and experimental performance assessment of today?s prominent robotic actuator technologies, that is, harmonic drives and quasi-direct-drives, is conducted in comparison with the MR clutching technology. Analytical models of five key performance metrics are developed for torque-to-mass, torque-to-inertia, backdriving loads, rendering stiffness, and power consumption. Finally, the design space of the three actuation technologies is drawn and performance potential in robotics is compared. Results show how MR actuators resolve a gearing conflict by decoupling the motor inertia through a fluidic interface, enabling gearing ratios between 50 to 100:1 with minimal output inertia, and thus, best in class accelerations. Results also show that MR actuators resolve a damping conflict by exploiting the serial positioning of the fluidic interface to tailor damping rates on demand, thus enabling rendering stiffnesses from null up to five times stiffer than harmonic drives. These unique dynamic characteristics combined with best-in-class torque densities in the 100 to 200 N.m/kg range, low backdriving torques, and low power consumption open the door to unseen robotic performance such as cobots capable of human-like haptic tasks.