Optimization of Merging Pedestrian Flows Based on Adaptive Dynamic Programming
Optimization of Merging Pedestrian Flows Based on Adaptive Dynamic Programming
复制标题
DOI：
10.23919/acc.2019.8814597
复制
发表时间：
2019-07
期刊：
影响因子：
--
通讯作者：
Chao Jiang;Yi Guo;Z. Ni;Haibo He
中科院分区：
文献类型：
--
作者：
Chao Jiang;Yi Guo;Z. Ni;Haibo He
人口密集区的行人流可能引发人群拥挤事故，有效的行人流调控是人流优化的重要手段。在本文中，我们研究的问题，调节两个合并的行人流，通过引入一个移动的机器人在流内移动。在行人集体运动过程中，通过动态的人机交互来调节行人流量。我们提出了一种基于自适应动态规划（ADP）的方法来学习最优的运动控制的机器人在真实的时间和行人流量最大化通过瓶颈区。使用行人运动的社会力量模型进行了广泛的模拟。仿真结果表明，我们提出的ADP控制的行人流量显着改善。
Pedestrian flows in densely-populated areas may cause crowd accidents, and effective pedestrian flow regulation is highly desirable for flow optimization. In this paper, we investigate the problem of regulating two merging pedestrian flows by introducing a mobile robot moving within the flow. The pedestrian flows are regulated through dynamic human-robot interaction during their collective motion. We propose a method based on adaptive dynamic programming (ADP) to learn the optimal motion control of the robot in real time and the pedestrian outflow through the bottleneck area is maximized. Extensive simulations are performed using social force models of pedestrian motion. Simulation results show that the pedestrian outflow is significantly improved with our proposed ADP control.