, Fmr Director, Intelligent Veh. Lab, Mechanical Engineering
In urban areas, Differential Global Positioning System (DGPS) devices suffer from problems with satellite view and with the reception of DGPS corrections over the cellular phone network. Moreover, the signals that do reach a GPS antenna suffer from multi-path interference, producing high-variance position estimates. These DGPS issues are addressed by fusing the position information provided by a vehicle positioning system (VPS) based on radio frequency identification (RFID) technology with data from scanning laser sensors. VPS provides "which lane" in the "lateral" direction, and a path distance from a known reference in the "longitudinal" direction. From this, the location of a bus in a particular lane can be computed. Position within the lane can be determined using a horizontal laser scanner mounted on the bus, which identifies the presence and location of curbs in order to determine the position of the bus in the lane with respect to the curb. Integration of VPS information, laser scanner information, and a "map" representation describing the optimal distance of the bus from the curb as a function of distance traveled along a lane, provides the lane information needed for driver assistance in a dedicated lane. This project produced a demonstration of a VPS-laser scanner-based positioning system capable of operating the IV Lab driver-assist system on an urban road in Minneapolis.