Building and navigating maps of road scenes using an active sensor

This paper presents algorithms for building maps of road scenes using an active range and rejectance sensor and for using the maps to traverse a porfion of the world already explored. Using an active sensor has some attractive advantages: It is independent of the illumination conditions, it does not require complex calibration in order to transform observed features to the vehicle's reference frame, and it provides 3 0 terrain models as well as road models. Using the m p built from sensor data facilitates navigation in two respects: The vehicle may navigate faster since less perception processing is necessary, and the vehicle m a y follow a more accurate path since the navigation system does not rely entirely upon inaccurate visual data. We presea a complete system that includes road following, map building, and map-based navigation using the ERIM laser range finder. We report on experimentationr of the system both on the CMU NAVL.43 and the Martin Marietta ALY.