Sobol sequence RRT* and numerical optimal joint algorithm-based automatic parking trajectory planning of four-wheel steering vehicles

Four-wheel steering can effectively improve turning agility and mitigate parking spatial requirement. Addressing the issues of low sampling point quality and poor efficiency in existing Rapidly-exploring Random Tree (RRT) and its improved algorithms for automatic parking assist (APA) system, a parking trajectory planning algorithm combining Sobol-RRT* with Reeds-Shepp curve and numerical optimal control within four-wheel steering kinematic model constraints is proposed in this paper to improve parking space utilization rate, cope with dynamic obstacles during parking process. First, the hierarchical framework of the proposed path planning algorithm is introduced, which is used as the basis of the planning algorithm, and the kinematics model of the four-wheel steering vehicle is established. Second, the pseudo-random sequences of RRT algorithm are replaced by Sobol sequences with uniform difference characteristics. Then, the parking trajectory planning problem is formulated with consideration of the system dynamic equation constraints based on the four-wheel steering kinematics model and the dynamic obstacle constraints based on the "triangular area method". Finally, the planned parking trajectory for the four-wheel steering vehicle is obtained by solving the optimal control problem and cubic spline curve fitting. Simulation in typical parking conditions validated the proposed planning algorithm on improvement of the APA system adaptability to challenging parking environment with dynamic obstacles.