Nonlinear disturbance observer-based compliance error compensation in robotic milling

Robotic machining is of growing interest around the world, but the joint deflection from the reference trajectory due to the robot's low rigidity restricts the use of robots only to machining with low cutting force. This paper proposes a new method for compensating the compliance error based on the nonlinear disturbance observer to reduce the deflection of robot joints against cutting forces. The disturbance observer designed in the joint spaces allows estimating the cutting forces online without any transformation between spaces so that the cutting forces could be obtained with high accuracy. The low-pass filter of the observer is optimized to reduce the effect of model uncertainty arising from the identification process. The deviation obtained by multiplying the joint compliances by the estimated cutting torques can be used to modify the reference position of robot joints. The proposed approach has the benefit to correct the robot's deflection in real time without a sensor. The experimental results reveal that the nonlinear disturbance observer can correctly estimate the cutting forces. Also, they show that the tool center point (TCP) deviation from the predefined path has been reduced using the proposed compliance error compensation. © 2022