Abstract: Aiming at the problems of large randomness, slow convergence speed and deviation of rapidly-exploring random tree algorithm, the basic fast random tree algorithm is improved by using cyclic alternating iteration search method to generate a new node and adopting a bidirectional random tree to do search simultaneously. A vehicle steering model is established to determine the constraint range of the vehicle steering angle, and the vehicle turning angle constraints are increased in the algorithm to reduce the deviation of the generated path, and improve the quality of the generated path. Node optimization is performed on the generated path to remove redundant nodes, shorten the path length, and improve the feasibility of the path. The B-spline curve is used to smooth the path by inserting the local end points, thus making the generated path more in line with the driving conditions of the vehicle. Simulation in MATLAB verifies the correctness of the algorithm.

Key words: intelligent vehicle;rapidly-exploring random tree;angle constraint;node optimization, path smoothing