自动驾驶汽车为何尚未就绪:问题、尝试性解决方案与未来方向的多学科综述
摘要
个人自动驾驶汽车是能够感知周围环境、规划路线并在很少或无需人类驾驶员参与下行驶的汽车、卡车和自行车。尽管业界近期取得了令人瞩目的技术成就,且领军企业家做出了充满希望的宣布,但迄今为止尚无任何个人车辆获准以“完全”或“半”自动驾驶模式(自动驾驶等级 4 和 5)上路行驶,且仍不清楚此类车辆何时才能成熟到获得此类批准。本综述采用综合与多学科方法来调查汽车领域面临的主要挑战,旨在识别仍困扰并延迟自动驾驶汽车商业化的问题。综述考察了当前技术的局限与风险以及研究者设计的最有前景的解决方案。这种负面评估方法并非出于悲观,而是出于提升对技术最先进水平、行业质量标准以及社会需求与期望的批判性认知的愿望。虽然本调查主要关注人工智能在感知与导航中的应用,但也旨在提供一幅扩大的图景,将纯技术方面与相关的以人为中心的方面联系起来,包括文化态度、概念假设和规范(伦理-法律)框架。审视更广阔的背景有助于凸显具有跨学科范围的问题,并识别可能受益于整体考量的解决方案。
引用
@article{arxiv.2311.09093,
title = {Why Autonomous Vehicles Are Not Ready Yet: A Multi-Disciplinary Review of Problems, Attempted Solutions, and Future Directions},
author = {Xingshuai Dong and Max Cappuccio and Hamad Al Jassmi and Fady Alnajjar and Essam Debie and Milad Ghasrikhouzani and Alessandro Lanteri and Ali Luqman and Tate McGregor and Oleksandra Molloy and Alice Plebe and Michael Regan and Dongmo Zhang},
journal= {arXiv preprint arXiv:2311.09093},
year = {2025}
}
备注
This manuscript extends the work "Applications of Computer Vision in Autonomous Vehicles: Methods, Challenges, and Future Directions." We have added several sections to explore autonomous vehicles from a multidisciplinary perspective. We propose changing the arXiv category to cs.RO, as the expanded content addresses broader autonomous vehicle topics aligning more closely with the Robotics domain