Research On Optimal Trajectory Planning And Motion Control Of Autonomous Racing Vehicle

This thesis aims to explore control potential of autonomous vehicles under extreme driving conditions,such as high-speed and rapid turning condition,by studying racing trajectory planning and motion control of autonomous racing vehicles(ARV).Three research topics,including time-optimal racing trajectory planning,high-precision motion control at ARC driving limits,scaled ARC test platform design and racing experiments,are addressed in order to tap the potential of the handling and racing ability of ARCs,which further provides theoretical supports for the dynamic control of autonomous vehicles at handling limits.(1)With regard to the coupling relationship between the racing path and vehicle speed,a simultaneous path and speed planning(SP~2)strategy is proposed.First,a time-optimal ARC trajectory planning problem considering constraints of track boundary and vehicle dynamics is formulated.Then,by designing a transition network connecting ARC steady action space and vehicle state,the time-optimal trajectory planning problem is transformed as a longest-distance optimal problems of each time step,which renders a shortest-time racing trajectory through receding horizon optimization.Simulation results show that the trajectory generated by the proposed SP~2strategy can save 20%of time compared to traditional centerline trajectory.(2)Considering the high computational burden of SP~2 strategy,an improved strategy based on viability theory,named SP~2-V,is proposed to efficiently plan the local racing trajectory.First,path points involving dangerous driving conditions,such as collisions and vehicle dynamics dissatisfaction,are deleted from the original state-action space using viability kernel training.A steady action space with high utility probability path points is offline selected,and a viability action set constrained by vehicle dynamics is formulated.Then,the time-optimal trajectory can be provided in a recursive planning way.Simulation results reveal that SP~2-V strategy can greatly improve the trajectory planning efficiency.(3)The third problem relates to highly precise ARC motion control.Two model predictive control(MPC)–based motion control strategies are proposed.First,a linear vehicle dynamic model based on model linearization is formulated,and a linear MPC(LMPC)controller is designed based on the model to achieve real-time calculation.Then,with consideration of vehicle load shift,the nonlinear MPC(NMPC)approach is exploited to realize motion control of the nonlinear ARC system.Results of simulations claim that the motion control precision of NMPC controller is better,while the LMPC controller is of higher potential to be applied in real time.(4)A scaled racing platform,including hardware components such as racing track,global positioning system and scaled ARC,and software components such as positioning,trajectory planning and motion control system of ARC,is designed and built.Then,a time-optimal trajectory planning and motion control experiment is carried out in the platform,which demonstrates that the strategy proposed in this thesis saves 24.7% racing time.