This paper proposes an algorithm to detect the individual obstacles on the road using u-v-stereo disparity histograms and a risk map for autonomous driving. The algorithm consists of three main parts. The first part involves separating the disparities corresponding to the road surface from the disparities relating to objects on a road image using v-lines generated from v-disparity histograms. Then a u-line corresponding to the boundary between the road surface and objects is detected using the v-line and u-disparity histogram. The third part involves detecting an individual obstacle by constructing a two-dimensional risk map. This approach is faster and simpler than road surface fitting approaches using disparities. In addition, this approach makes path generation for autonomous driving efficient because it provides geometric information regarding individual obstacles. Experimental results show that the proposed algorithm detects obstacles with 98% precision in a variety of environments.