| Path planning is one of the most important parts of the vehicle navigation system.With the rapid increase of geographic information data and car ownership in China, thetransportation system becomes more and more complex, which makes it quitechallenging to meet the requirement of individualized demand of drivers. Therefore,developing the path planning algorithm under traffic restriction to meet therequirement of the vehicle navigation system has become a critical problem that needsto be solved. In this dissertation, the algorithms of transportation-forecast optimumpath planning, area-through path planning, and target guiding under the restraint ofspecial road net were proposed.A computation model that can evaluate whether the number of the floating car canmeet the requirement of the dynamic path planning was initially developed. Thethreshold control method was adopted to filter the received dynamic traffic informationon the premise that the accuracy can meet the requirement. At the same time, themethods of weighted average and exponential smoothing were first used to supplementthe missing data, then using the data from the adjacent road segment, together withtopo characteristics of the road net, to further complete the missing data. The noisedata was repaired using the method of reconstruction of the principal components. Theclustering analysis method was applied to the repaired traffic data to build thematching template, and short-term forecast on the traffic data was performed using themethods of multi-factor pattern matching and adaptive correction based on theforecasting error. The traffic model of the urban road net was then developed, whichmakes it possible to develop the time optimal path planning based on forecast.Compared with the original static path planning, the optimum vehicular pathplanning with traffic forecast can take the dynamic traffic restriction into consideration,thus it enables the driver travels along the optimum route avoiding the traffic jam. Inorder to rapidly read and update the dynamic traffic data, the dynamic traffic databasewas established in this dissertation, which can provide convenient basic support for thedynamic updates of the electronic map. The optimum path planning with traffic forecast can fulfill the dynamic path planning of the plane road net through calculatingthe cost between two nodes using traffic model of the urban road net and computingthe waiting time at the cross.The area-through path planning proposes a new method of path planning. Theusers can select the passing area according to their own needs. As a result of that, pathplanning under area restriction was realized. The algorithm proposed in thisdissertation develops the model of area-through path planning, enabling the results topass the designated area. The area-through path planning was fulfilled through twosteps: the virtual target path planning and the direction-changing path planning. Thevirtual target path planning is achieved by leading the car to pass the designated areaby setting the virtual-target point and using A*algorithm, while the direction-changingpath planning can lead the car to pass the target area through changing the pathplanning heuristic cost.Based on the basis of the spider-web road network the target guiding pathplanning restrained under the special road net preprocesses the road net data in block.It estimates whether the directional joint-node is on the optimum path to this blockthrough calculating the cost from other nodes to boundary nodes of the block. Thetarget guiding path planning only topo the nodes with optimal flags in the restraint area,thus it can reduce the topological nodes and increase the speed of path planning. |
PDF Full Text Request