Simulation of sliding mode impedance control for a 6 DOF anthropomorphic robot

Robust control is a way of dealing with model uncertainties that can render the controller less sensitive to modelling errors and disturbances. Herein, the properties of sliding mode control are applied to impedance control to form a robust impedance controller. The proposed controller consists of two parts: a nominal part based on a nominal model of the robot and a robustness part compensating for the difference between model and real robot. Different kinds of robustness parts give differences in chattering, steady-state error and complexity. We can control within certain accuracy, if the uncertainty does not exceed certain upper bounds. The sliding mode impedance controllers were verified by simulations and compared to the old computed torque controller. The parameter uncertainty ranges from 0.5 to 2. Simulations show that the sliding mode controllers have a small impedance error. For these uncertainties the computed torque controller has large errors, or becomes unstable. Some qualitative tests have been carried out showing the control laws have been implemented correctly. However, quantitative experiments need to be done to validate the control laws. Furthermore, we recommend to make the boundary layer width adaptive, to always have the highest accuracy possible while keeping robustness.