【摘要】 针对广义速度障碍(Generalized Velocity Obstacles, GVO)算法解决动态环境避障问题时计算量过大、动作过于保守等问题,提出了速度障碍法(Velocity Obstacles, VO)与GVO组成的分级避障控制生成策略,使用VO判断所选控制生成的速度矢量是否存在碰撞危险,如果存在碰撞危险则进一步使用GVO对该控制进行判断,否则认为该控制为安全控制。通过建立碰撞时间评价函数和目标适应度评价函数选取最优控制。当可达控制集中所有控制都不可行时,通过可达控制集外的一组控制来启发当前控制的生成。仿真结果表明,所提算法有效的避免了GVO算法中不必要的计算,在不同场景中至少缩短算法运行时间15.91%,改善了GVO算法无解时避障,得到了更合理的轨迹。更多还原

【Abstract】 Aiming at the problem that the Generalized Velocity Obstacles(GVO) algorithm has too much computation and too conservative action when solving the problem of dynamic environment obstacle avoidance, a hierarchical obstacle avoidance control generation strategy composed of Velocity Obstacles(VO) and GVO is proposed. Firstly, VO was used to determine whether there is a collision risk in the velocity vector generated by the selected control. If there was a collision risk, GVO would be used to judge the control, otherwise, it was considered to be a safety control. The collision time evaluation function and the target fitness evaluation function were established to select the optimal control. When all the controls available were not feasible, another group of outside control would be used to inspire the generate of current control. The simulation results show that the proposed algorithm can effectively avoid unnecessary calculations in the GVO algorithm and shorten the algorithm running time by at least 15.91% in different scenes, and the obstacle avoidance of GVO is improved by this new system. Especially, obstacle-avoidance is improved when GVO algorithm has no solution, and a more reasonable trajectory is obtained.更多还原