Obstacle detection is one of the key points of unmanned vehicle research, which needs to obtain obstacle data from multiple angles. A method on driving obstacle detection of new energy vehicle unmanned vehicle based on infrared technology and lidar is proposed. Using Trimble MX2 3D lidar equipment and infrared sensor, the obstacle data during driving is obtained, and the data acquisition error is reduced through preprocessing. The processed obstacle data points are connected into an obstacle contour, and the obstacle detection is completed according to the obstacle classification information. Based on the building of a detection performance test environment,, experimental results show that the combination of infrared technology and lidar can accurately detect all obstacles, accurately divide the types of obstacles, keep the misclassification rate of driving obstacle classification between 0%~10%, and keep the detection time within 2 s. The detection and recognition effect is good, which can improve the safety factor of unmanned vehicle driving at night.