Object Detection Using Vision and LiDAR Sensor Fusion for Multi-channel V2X System

In this paper, we propose an algorithm that recognizes the environment around the vehicle using Vision and LiDAR sensor fusion, and extracts the object affecting the multichannel V2X communication system and transmits information to the TCU. A general object detection algorithm for autonomous vehicle detects pedestrians, signs, vehicles. However, in this paper, we detect the surroundings that affect the communication performance. For example, the detection targets are not only a bulky vehicle such as a large truck or bus but also a bulky structure such as a building, a tunnel or a soundproof wall. The proposed method has the three steps. The first is receiving cameras and LiDAR sensors and to preprocess the data from each sensor. In this step, the distortion of image data is calibrated, and the LiDAR sensor separates the ground point group and the non-ground point group. The second step is sensor fusion. The coordinate system is integrated into a transformation matrix between two predetermined sensors, and the results of the two data are fused. The last step is object classification using CNN. The proposed method can improve the reliability and stability of communication by detecting the objects affecting the communication performance in advance and changing the channel parameters of the multichannel TCU.