Aiming at the problem of depth information loss in target recognition and position estimation of monocular mobile robots,an improved Kalman filter algorithm combining radar ranging information and azimuth was proposed to improve the positioning accuracy of lidar assisted distance measurement.The YOLO network was used for target recognition,and then basic computer vision calibration methods were used to obtain target directional angle information.Radar was used for multiple directional ranging to obtain multiple observation values.Then,during the target position calculation,different confidence levels were weighted according to the distance and azimuth information of different observations to change the observation noise and system noise of the Kalman filter,and then the Kalman gain was dynamically improved to achieve adaptive target position calculation.The simulation and experimental results show that this method can achieve more accurate positioning compared to relying solely on radar for ranging supplementation,and has good application prospects.