Research on Unmanned Motion Planning Algorithm for Urban Roads Based on Multiple Condition Screening
Yizhuo Liu, Liqiang Liu, Xiaohang Yu · 2020
With the rapid development of driverless driving technology, higher requirements have been placed on the motion planning algorithms of unmanned vehicles. This paper proposes an unmanned motion planning algorithm for urban roads based on multiple condition screening. This algorithm screens a part of the trajectories through the initial condition screening, then performs cost function calculation and ranking on the remaining trajectories, and add them to the set of alternative trajectories. The purpose is to improve the calculation efficiency of the subsequent cost function. Considering that the cost function and the proportional coefficient of the traditional algorithm for obstacle avoidance require a lot of manual experiment adjustment and selection and may miss the optimal solution. This paper cancels the use of the obstacle avoidance cost function and performs obstacle collision detection and screening in the alternative trajectories set. The purpose is to obtain the optimal trajectory in the set without collision under the same sampling density. Simulation results show that the algorithm can achieve the task of driverless vehicle motion planning on urban roads. Compared with the traditional algorithm, the algorithm is more efficient, avoids manual adjustment of parameters, and is closer to the optimal solution with higher reliability.