Research On Autonomous Vehicle Motion Planning Method Using Parameterized RRT Based On Guiding Area

Autonomous vehicle can not only reduce the incidence of traffic accidents,but also can improve the efficiency of car travel.Therefore,it has received widespread concern from scientific research institutions and enterprises.The motion planning technology is one of the core technologies in autonomous vehicle and has always been the reaserch hot and difficulty.The motion planning of autonomous vehicle should consider the following two aspects.On the one hand,autonomous vehicle needs real-time planning to meet the vehicle kinematic constraints of the feasible path so that it can cope with rapidly changing driving environment.On the other hand,the motion planning algorithm should have good universality,which can be carried out in many different driving scenes successfully.But the existing solutions still can not solve the above problems perfectly.Accordingly,the motion planning algorithm using parameterized RRT(Rapidly-exploring Random Tree)based on the guiding area is provided.The detailed researching contents are as follows:1)Aiming at the problem that the path generated by the RRT algorithm does not satisfy the feasibility of the autonomous vehicle,this dissertation proposes a parameterized node generating method,which ensures that the resulting path can always satisfy the conditions of the Bezier curve parameterization,hence it can make sure that the path planned by the RRT algorithm is curvature continuous and satisfies the feasibility constraint of the autonomous vehicle.2)In order to reduce the planning time of the RRT algorithm and improve the real?time performance of the algorithm,this dissertation proposes a goal tree strategy and the new termination checking method.On the one hand,by extending the target point into a goal tree with a certain number of nodes,the probability of the random tree finding the feasible path is improved.On the other hand,the new termination checking method tries to connect with the goal tree to detect the feasibility of the connection when a new node is created,so as to further improve the planning efficiency of the algorithm.3)Aiming at the problem that the RRT algorithm has large blindness in node sampling and poor resulting path in the aspect of length,this dissertation proposes a guiding area mechanism.The A*algorithm is used to generate a guiding area in a low-resolution grid map,and then limit the RRT to sample only in the guiding area,so the RRT's sampling space is optimized and the quality of the path generated by the RRT algorithm is improved.Finally,the superiority,validity and practicability of the motion planning algorithm proposed in this dissertation are verified by comparing simulation and real vehicle experiment.