Autonomous driving solutions stretch over different disciplines and technologies e.g., sensors, communication, computation, machine learning, data analytic, etc., that need to be smartly stitched together for achieving end-to-end solutions. In this paper, we discuss the vision of level 5 autonomous vehicles (AV), relevant challenges, and analysis of the research literature. The paper focuses on the role of communication for connected and automated vehicles. Furthermore, the need for implanting intelligence in different architectural components and for various Autonomous Driving relevant operations is discussed. Challenges specific to communication, perception, service orchestration, service mobility, etc. are highlighted, relevant research literature analyzed, and potential solutions sketch is provided. We have also provided an overview of the large-scale proof-of-concepts around the globe that guide the readers towards studying different aspects of the autonomous driving and perspectives of stakeholders therein. The potential of Satellite-Air-Ground-Integrated-Networks (SAGINs) is studied for realizing the objectives of envisioned higher level autonomous driving. Based on the exhaustive analysis of the research work, this work concludes that there is a need for zooming out strategy, where the novel architectural, technological, and AI-based solution approaches are crafted by capturing the end-to-end system with the focus on most (if not all) stakeholders and their objectives.

自动驾驶解决方案跨越不同的学科和技术，例如传感器、通信、计算、机器学习、数据分析等，需要巧妙地组合在一起以实现端到端解决方案。在本文中，我们讨论了 5 级自动驾驶汽车 (AV) 的愿景、相关挑战以及对研究文献的分析。本文重点关注通信在联网和自动驾驶汽车中的作用。此外，还讨论了在不同架构组件中植入智能以及各种自动驾驶相关操作的需求。突出了通信、感知、服务编排、服务移动性等方面的挑战，分析了相关研究文献，并提供了潜在的解决方案草图。我们还概述了全球范围内的大规模概念验证，引导读者研究自动驾驶的不同方面以及其中利益相关者的观点。研究了卫星-空中-地面-综合网络 (SAGIN) 的潜力，以实现设想的更高级别自动驾驶的目标。基于对研究工作的详尽分析，本工作得出结论，需要缩小战略，通过捕获端到端系统，重点关注大多数（如果不是全部）利益相关者及其目标，制定新颖的架构、技术和基于 AI 的解决方案方法。