Real and effective detection of the drivable area on roads is essential for the path planning, real-time navigation, and obstacle avoidance of unmanned sweeping machines. After obtaining the laser point cloud of the road environment using 3D LiDAR, an improved ground segmentation algorithm is first used to segment the ground and non-ground point cloud. Then, for structured roads, candidate road boundary points and obstacles are determined based on the geometric features of the point cloud. The random sampling consistency algorithm is combined with the least squares method to extract the curb boundary lines and the isolation lines that remove obstacles on the road surface. For unstructured roads, laser reflectivity clustering is used to extract the road surface, and the sliding window method is used to determine the boundary points and extract the boundary curves using B-spline. In turn, by integrating the boundary lines obtained from both algorithms using distance criteria, the drivable area of the road is obtained. Finally, an unmanned driving experimental system is used to extract drivable areas. The experimental results show that the fusion algorithm proposed achieves an accuracy of 96. 5% and a recall rate of 92. 7% for extracting the drivable area of laser point cloud data on mixed roads, with an average processing time of only 29 milliseconds. The accuracy of measuring the width of the real drivable area reaches 97. 1% , proving that the road drivable area fusion algorithm offers both high accuracy and efficiency.