Human-Robot Cooperation for Autonomous Vehicles and Human Drivers: Challenges and Solutions
Human-Robot Cooperation for Autonomous Vehicles and Human Drivers: Challenges and Solutions
复制标题
自动驾驶汽车和人类驾驶员的人机合作：挑战和解决方案
DOI：
10.1109/mcom.001.2001241
复制
发表时间：
2021
影响因子：
11.2
通讯作者：
R. Rajkumar
中科院分区：
文献类型：
--
作者：
Shunsuke Aoki;Chung;R. Rajkumar
联网和自动驾驶车辆 (CAV) 变得越来越可行，这些车辆有望提高道路安全、交通吞吐量、能源效率和人类舒适度，特别是在道路交叉口以及变道、并道和无保护转弯操作时。然而，在人类驾驶车辆完全被自动驾驶车辆取代之前，可能需要很长的过渡期。为了在公共道路上利用自动驾驶技术，我们必须为混合交通环境设计和开发安全的交互与协作，在这种环境中，人类驾驶和自动驾驶车辆相互协作，以避免车辆碰撞和可能的僵局。在本文中，我们研究了自动驾驶汽车和人类驾驶员安全合作的要求和挑战，并回顾了三个未来的组成部分：道路基础设施、车辆通信以及在线识别和预测。首先，道路基础设施有助于提高 CAV 作为集中协调器进行安全合作的能力。此外，车辆间通信可能是以去中心化方式安全合作的关键，可以保证多辆联网车辆的道路安全。在线运动预测和实时手势识别对于以去中心化方式与非联网车辆开发安全协作系统也至关重要。这三种方法并不排斥，CAV 在实践中可以互补使用它们。最后，我们讨论每种方法的可行性和局限性，并总结未来的研究方向。
Connected and automated vehicles (CAVs) are becoming increasingly feasible, and these vehicles are expected to enhance road safety, traffic throughput, energy efficiency, and human comfort, especially at road intersections and for lane changing, merging, and unprotected turning maneuvers. However, there is likely to be a long transition period before human-driven vehicles are completely replaced by such automated vehicles. To leverage the autonomous driving technologies on public roads, we have to design and develop safe interaction and cooperation for mixed traffic environments, where human-driven and autonomous vehicles cooperate with one another in order to avoid vehicle collisions and possible deadlocks. In this article, we study the requirements and challenges for such safe cooperation of autonomous vehicles and human drivers, and review three prospective components: road infrastructure, vehicular communications, and online recognition and prediction. First, road infrastructure contributes to improving CAV capabilities for safe cooperation as a centralized coordinator. In addition, vehicle-to-vehicle communications might be key to safe cooperation in a decentralized manner that can guarantee road safety for multiple connected vehicles. Online motion prediction and real-time gesture recognition are also essential to develop a safe cooperation system in a decentralized manner with non-connected vehicles. These three approaches are not exclusive, and CAVs may complementarily use them in practice. Finally, we discuss the feasibility and limitations of each approach and conclude with future research directions.