A Realtime Quadratic Sliding Mode Filter for Removing Noise

This paper presents a sliding mode filter that effectively removes impulsive noise and high-frequency noise with producing much smaller phase lag than linear filters. It is less prone to overshoot than previous sliding mode filters and it does not produce chattering. In addition, it is computationally inexpensive, and thus suitable for realtime applications. The proposed sliding mode filter employs a quadratic surface as its sliding surface, which is designed so that the output converges to the input in finite time if the input value is constant. Its algorithm is derived by using the backward Euler discretization, which prevents chattering. The effectiveness of the filter was shown by experiments by using an ultrasonic sensor and an optical encoder.