Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : June 06, 2021 - June 08, 2021
This article proposes a method to improve the operability and stability of collaborative work between a human and an impedance-controlled robot by adaptively changing the impedance characteristics online. In the proposed method, the impedance parameters of the collaborative robot are varying according to the stiffness of the human arm and the speed of the object to be manipulated. Specifically, a normalized stiffness of the human arm during collaborative work is estimated using a recurrent neural network (RNN) from the myoelectric potential signals, and a variable impedance control system that changes the impedance characteristics according to the estimated value is implemented to improve the stability. Furthermore, changing the impedance parameters by additionally using speed of the manipulated object is also proposed to keep the operability nimble. The experimental results showed that the proposed variable impedance control achieved the same nimble operability as the light impedance model while maintaining the stability, and it has been confirmed that human operator consumes less power for the task by the proposed adaptation depending on the object’s speed.
