Multi-sensor Fusion Tracking Algorithm by Square Root Cubature Kalman Filter for Intelligent Vehicle

Abstract

With the development of unmanned driving technology, the demand for tracking accuracy is increasing. To rely on a single sensor to obtain detection results in complex environments is limited in accuracy, and multi-sensor fusion is an effective method. Therefore, the environmental sensing technology based on multi-sensor fusion is one of the present research hotspots. This paper proposes a multi-sensor fusion tracking algorithm based on the square root Cubature Kalman filter (SRCKF) for the purpose of nonlinearity of the vehicle target tracking system. This method establishes the equation of state and the measurement equation using the dynamic model, and utilizes the multi sensor measurement signal. A data fusion method based on cubature Kalman filter (CKF) with nonlinear system to avoid errors caused by linearization of nonlinear systems by extended Kalman filter (EKF). In the filtering process, the covariance square root matrix is used in place of the covariance matrix participating in the iterative operation. It effectively avoids the divergence of the filter and improves the convergence speed and stability of the filtering algorithm.