An Adaptive Kalman Filtering Algorithm Without Using Kinematic Models

Abstract

The performance and accuracy of Kalman filter depends on its gain value related to the process noise covariance and the measurement noise variance which may vary according to experimental settings such as noise and sampling time. Thus, setting the appropriate values for the noise variances that fit for a wide range of experimental setting is a challenge for conventional Kalman filter. This paper proposes an adaptive Kalman filter with the adaptive noise variance for velocity estimation without using kinematic model. By applying only the quantized position measurement signal generated from the optical incremental encoder, an adaptive process noise variance is proposed. The experimental results show that the proposed method outperforms the conventional Kalman filter in achieving accurate and smooth velocity estimation without large time delay.