Autoregressive model with Kalman filter for estimation of physiological tremor in surgical robotic applications

In real-time implementation computational complexity plays vital role. This paper focuses on adaptive signal processing of physiological hand tremor for tremor cancellation in robotic devices. The physiological tremor is modelled with AR(3) process that has less computational complexity compared to other model based existing methods. In this paper, filter coefficients are updated with Kalman filter to improve the performance. The existing method AR-LMS and the improved method AR-Kalman are implemented in real-time for tremor compensation. A comparative study is conducted on the algorithms with the tremor data from microsurgeons and novice subjects. Experimental results shows that the proposed method AR with Kalman filter improves the accuracy by at least 10% in real-time compared to AR with LMS.