タスク優先順位による多目的制御問題の分割を利用した全方向移動車両の任意形状障害物回避

抄録

Omni-directional vehicles have large degrees of freedom and can move flexibly, but this flexibility also increases the complexity of motion planning. To exploit the performance of the vehicle, multi-objective optimization is a powerful approach. However, minimizing a scalar evaluation function often leads to undesirable behavior due to local minima, and it is not easy to balance the priorities between tasks by adjusting the weight parameters. This study presents an approach to solve a multi-objective optimization problem by dividing it into hierarchical subproblems based on task priorities. The quality of the solution is improved by (1) narrowing down the search space reasonably and (2) combining gradient-based and sampling-based optimization methods. In the motion planning for navigating through obstacles, the proposed method obtained solutions that ensure reaching the goal more accurately compared to the conventional method (MPPI).