Research On GIS Database Construction Andapplication For UGV In The Urban Traffic Environment

As the research focus of Intelligent Transportation System, Unmanned GroundVehicles (UGVs) will reduce traffic accidents and improve traffic congestion, thusimproving the efficiency of the whole urban traffic system by its high intelligence. UGVsuse on-board sensors to perceive the environment around the vehicle and obtaininformation of roads, current locations and obstacles to control steering and speed of thevehicle, so that the vehicles can be safe and reliable on the road. The increasingly complexurban traffic environment makes UGVs perceive much more traffic information, which willno wonder increase the perceptive difficulty for UGVs. On the other hand, GeographicInformation System (GIS), which has strong capabilities of storage and management forgeographic information, will play a key role in environment perception for UGVs.Based on the purpose of assisting UGVs in driving autonomously in the urban trafficenvironment, this dissertation presents a method for the construction of GIS database whichconsiders the perceived demand of UGVs in the urban traffic environment. Moreover, theapplicability and effectiveness of the GIS database will be verified by the algorithms ofmap matching and path planning for UGVs. The mainly studying results of this dissertationare as follows:Firstly, the perceived needs of UGVs are analyzed and the road network model is builtin the urban traffic environment. The above work makes contribution to construct GISdatabase for UGVs, which stores traffic information related to the perception anddecision-making of UGVs.Secondly, an algorithm of map matching based on graphics is presented, which canmatch the real-time location of UGVs on the map of GIS precisely, and the surroundingtraffic information will be provided by the GIS database.Finally, a prediction model of travel time based on RBF neural network is establishedto predict the travel time of each road section. Then, the actual travel time achieved by realcar experiments is used to constantly update the predictive travel time on the basis ofBayesian theory in order to improve the predictive accuracy. After the above work, theleast-time path planning algorithm based on the GIS database and the path selection methodbased on probability are proposed, and the simulations are conducted to verify theeffectiveness of the algorithms and the applicability of GIS database.Simulations have shown that the proposed GIS database performs well in assisting UGV to move autonomously in the urban traffic environment.