Design And Implementation Of An Emotional Learning Controller For Force Control Of A Robotic Laparoscopic Instrument
Purpose: Force control of robotic instruments is a difficult task due to the uncertainties caused by changes in the instrument’s geometrical and mechanical characteristics during surgery as well as the nonlinear dynamics of the instrument. A new approach based on an intelligent controller is developed to control the force interactions of a robotic surgical instrument with delicate soft tissues. This feature assists the surgeon by providing a safe grasp of soft tissues during dissection or suturing. Besides, by controlling and optimizing the magnitude of the instrument/tissue contact forces, controlled grasp will significantly reduce the surgery trauma.
Method: The controller is devised using a neuro-fuzzy regulator that receives the tracking error and its derivative as inputs, and a PD critic that evaluates the actual pinch force and produces an emotional signal. The controller tunes its parameters by means of minimizing the critic’s output signal, i.e., stress, so that the force tracking error is reduced. Numerical simulations and experimental tests were performed to evaluate the controller.
Results: Simulation tests revealed that the controller can effectively adapt its rules when the instrument’s geometry and frictional behavior changes. The experiments revealed a settling time of 0.7 s with 3.1% overshoot. In comparison with a PID, the proposed controller reduced the mean squared error (MSE) by 94% for a target constant force, and 24% for a target sinusoidal trajectory.
Conclusion: the proposed controller showed a superior performance in force control of tissue in safe grasp in comparison with a PID particularly for constant target forces.