The safety traffic of intellectual vehicles is mainly carried out through environmental sensory perceptual system. This thesis uses the data collected by the vehicle-mounted laser radar and puts forward to a kind of improved straight-line feature pick-up algorithm SEF-LSM-BO. It experiments on the four kinds of algorithm using the same frame of laser radar data. It turns out that compared with SEF, LSM and SEF-LSM; SEF-LSM-BO can quickly and accurately detect and merge boundaries, mark surrounding obstacles and provide traversable areas.