Vision/Laser Sensor Fusion for Target Tracking and Car Following Control in Complex Environment and Driving Conditions

This paper presents an economical and reliable vision/laser-based car following control system. The vision sensor is integrated with a 2-D laser scanner to achieve reliable moving target recognition, discrimination and tracking performance. In highly complex environment and dynamic driving conditions, where the targeted vehicle can hardly be distinguished from nearby vehicles solely based on laser scanner data, the vision sensor can provide more reliable information of the target. Through the Consensus- based Matching and Tracking of Keypoints (CMTK) algorithm, the vision sensor is able to assist the laser sensor in determining the location of the target vehicle in complex environment. Thereby, the car following control (CFC) system is developed with the similar concept of the existing adaptive cruise control (ACC) system, which is typically designed for mid to high speed driving conditions. Yet, the proposed CFC system is aimed for low to mid speed driving with frequent stop-and-go modes. Moreover, the steering control is integrated with brake/speed control to enable car following in all driving and maneuvering conditions.