This paper discusses the research conducted at CMU in several topic areas: increasing the dynamic range of current cameras to better handle outdoor scenes, detecting roads in heavily shadowed scenes and in rapidly changing illumination, and identifying intersections. This research is necessary to add new capabilities to our road following vision system. We built a new system, using the results from our research, that could identify roads and intersections in shaded scenes. We also transferred some of our algorithms to the WARP, an experimental supercomputer developed at CMU, to test the capabilities of current parallel machines.
Transactions of JWRI 17 (1), 155-168, 1988-05
Osaka University