高风险驾驶情景中的高级纵向控制与碰撞规避的深度强化学习
微分几何
2025-10-07 v3 数学物理
代数几何
math.MP
辛几何
摘要
现有的高级车辆辅助系统(ADAS)主要关注前方车辆,常常忽视后随车辆的潜在风险。这一疏忽可能导致在高风险情景(如高速、紧凑、多车辆情景)中处理无效,因为一辆车的紧急制动可能引发连锁碰撞。为克服这些限制,本研究引入一种新颖的深度强化学习算法用于纵向控制和碰撞规避。该算法有效考虑了前导车辆和后随车辆的行为。在密集交通中的高风险情景模拟实施中,该算法 demonstrated ability to prevent potential pile up collisions, including those involving heavy duty vehicles.
引用
@article{arxiv.2404.19088,
title = {A Geometric Realization of Spherical T-Duality via $\star$-Diagrams},
author = {Leonardo F. Cavenaghi and Lino Grama and Ludmil Katzarkov},
journal= {arXiv preprint arXiv:2404.19088},
year = {2025}
}
备注
This is the version to be submitted. It is more specialized. The remaining from other versions and HMS-related questions shall appear elsewhere, jointly with other authors