External torque sensing algorithm for flexible-joint robot based on disturbance observer structure

Since a flexible-joint robot (FJR) consists of two subsystems centered around joint-torque (J-T) sensors, there are two independent resultant torques, i.e., an actuating motor torque and an external link torque, applied to the system. In this paper, an external torque sensing algorithm for the FJR is proposed to estimate both torques simultaneously by using the disturbance observing property of the disturbance observer (DOB). The proposed algorithm has no restriction of selecting Q-filter such that a high-performance low-pass filter can be applied for accurate estimation. The estimated torques can be very useful for FJR applications. As an illustrative example, the estimated actuating motor torque containing the motor disturbance is utilized to the motor disturbance compensation of the FJR with modified Q-filter due to the compensation feedback loop. The basic structure of the proposed algorithm is illustrated, and the performance is verified though a custom-designed experimental testbed for a vertical configuration with a gravitational load.