In order to enhance the real-time and accuracy of the control of automobile adaptive headlights under multi-type light source road conditions at night.In this paper

based on the capture of the coordinates of the position of the moving light source in front by the vehicle camera

a prediction method embedding an adaptive Kalman filtering（KF）algorithm in the AR model is proposed.The AR model can make real-time predictions of the movement trajectory of the moving light source at night in multiple time steps.The adaptive modified KF algorithm performs real-time discrimination on the deviation of the historical trajectory data of the moving light source in the AR model according to the current driving environment characteristics of the vehicle Amendments to ensure the accuracy of model predictions.Simulation results show that the prediction algorithm has extremely high accuracy and real-time performance

and can effectively improve the control performance of the adaptive headlight system.