2019 IEEE International Conference on Robotics and Biomimetics (ROBIO) | 2019

Joint Torque Estimation for Dynamic Jumping Control of Compliant Gear-Driven Legged Robot*

 
 
 
 
 

Abstract


This paper is to explain compliant torque control for gear driven legged robot. In this paper, active torque compliance control is realized without torque sensor or any mechanically elastic component. The joint torque estimation is very important for torque control. However, Joint torque sensor for medium-size legged robot is so large and heavy that it increases the inertia of the leg, slows down the dynamic response, and leads to complicated electrical system. Therefore, we only use the encoder information and the motor current to estimate the joint torque without torque sensors. Compliant torque control have advantages of absorbing impacts and saving energy. But, torque control is difficult to guarantee the position precision, because the accurate model of the robot is hard to obtain. In order to ensure both compliance and position accuracy, hybrid torque and position control is applied to our robot. In the jumping motion experiments, it is verified that the joint torque can reach peck value in 4.75 ms when jumping up. While The landing peak torque in hybrid control is lower than 1/6 of that under position control and 1/2 of that in torque control. This indicates that the compliance torque control can absorb a large part of shock.

Volume None
Pages 95-101
DOI 10.1109/ROBIO49542.2019.8961584
Language English
Journal 2019 IEEE International Conference on Robotics and Biomimetics (ROBIO)

Full Text