Autonomous Vehicle Simulation System, Off-road Environment

The control module is one of the most important modules of Autonomous Land Vehicle(ALV). Before actual experiments, the usage of simulation for control strategy is helpful inreducing the risk and speeding the period of research. The calculating method is mainly used inthe simulation of dynamics system before, the result of which isn't clear enough visually, andcan't fully incarnate the movement of the object. The application of virtual reality (VR)technique in the simulation of dynamics system offers a new approach of simulation, and has animportant meaning for improving the simulation effect. This method uses data computed fromdynamics model to activate the movement of geometry model. Accordingly it can simulate themovement of objects in the virtual scene and displays the result of simulation in an easyunderstanding way. This thesis aims at research works about how to construct a virtualsimulationsystemofALVtheoreticallyandpractically.At first an introduction of the development of ALV and vehicle simulation technologybased on virtual reality (VR) is made. In chapter 2, the basic concept and usage of softwareenvironment, based on the real-time 3D modeling tool of Creator and the applicationdevelopment software of Vega, are introduced. In chapter 3, three steps are described in theexploitation of the scene simulation system. The first one is to build 3D models byCreator. ThesecondoneistoconfigurethevirtuesceneinLynx.ThelastoneistodrivethesceneusingVegaand VC, and then a Vega simulation system is established based on MFC. In chapter 4, theschemeof buildingdynamicssimulationmodel andthedataintercommunionbetweenthemodelandvirtualenvironment arediscussed. Inthisthesis,themodelisconstructedbyrealdata,whichare gathered from experiments of ALV, and the model is imported into the simulation system tocompute and update the states of ALV. At last, a virtual steeringwheel used to control the statesofALVisconnectedtothesystemthroughserial interface.Theresultsofthesimulationindicatethatthismethodishelpfulinconfirmationandimprovementofvehicledynamicsmodel.