Abstract: To obtain the information of drivable area and obstacles for a driverless car, unique characteristics of the road edge data points were summarized and concluded by analyzing a large number of scanning lidar data, and a road edge detection algorithm was proposed based on the features of road edge data points and multi-layer fusion technology. DSm T was applied to establish a grid map for the road environment in front of the unmanned vehicle. The DSm T conflict coefficient was used to detect dynamic obstacles. Finally, the clustering and information extraction of dynamic obstacles was completed by the expansion algorithm, erosion algorithm, and the improved eight neighborhood labeling algorithm.Resultsshow that the algorithm can stably and accurately perceive the environment information around driverless vehicle.