Estimation of tire road friction coefficient based on image recognition and FC-UKF

In order to accurately estimate the tire road friction coefficient (TRFC), this paper presents a fusion estimation method based on image recognition and vehicle dynamic information. Firstly, TRFC is obtained by the onboard camera, and the convolutional network is used to recognize and classify the road surface type; Building on this foundation, a fast-converging unscented Kalman filter (FC-UKF) is devised. It adaptively adjusts the noise covariance matrix by leveraging the principle of minimum error entropy (MEE). Finally, a spatiotemporal synchronization estimation rule for image recognition and dynamic information fusion estimation is established. The simulation results show that the proposed fusion estimation method can accurately obtain the tire-road adhesion coefficient, compared with a single estimation method, it effectively improves the convergence speed and estimation accuracy of the estimation.