自动驾驶汽车在交接过程中的流畅和低延迟视频流
Smooth and Low Latency Video Streaming for Autonomous Cars during Handover
摘要：
自动驾驶汽车有望带来许多好处，其中包括提高交通效率和可靠性，减少燃料消耗，这将产生巨大的经济和环境影响。这项技术的成功在很大程度上依赖于其周围实体的全面态势感知。只有当一切都联网时，包括车辆，用户和基础设施，并在附近的物体之间交换感知数据以提高它们的感知能力，才能实现这一目标。然而，无论如何，在回路中仍然需要人为干预来处理看不见的情况或补偿不准确或不适当的车辆决定。在这种情况下，除了LIDAR等其他数据外，视频馈送被认为是为人类提供真实情况以最终做出正确决策的关键。然而，如果视频没有及时交付，它就会变得毫无用处，或者可能产生灾难性的后果。另外，流式传输的视频中的任何中断（例如，在穿越国家间边界时的切换操作期间）对用户来说是非常烦人的，并且也可能造成损害。在本文中，我们首先描述两个重要的用例，即远程驾驶和排队，其中视频的及时交付非常重要[1]。此后，我们详细介绍了我们实施的解决方案，以适应上述自动驾驶车辆的用例。通过在本地和LTE网络中的广泛实验，我们证明了我们的解决方案可以确保非常低的端到端延迟。此外，我们表明，我们的解决方案保持视频中断尽可能低的切换操作过程中。
英文摘要：
Self-driving vehicles are expected to bring many benefits among which enhancing traffic efficiency and relia-bility, and reducing fuel consumption which would have a great economical and environmental impact. The success of this technology heavily relies on the full situational awareness of its surrounding entities. This is achievable only when everything is networked, including vehicles, users and infrastructure, and exchange the sensed data among the nearby objects to increase their awareness. Nevertheless, human intervention is still needed in the loop anyway to deal with unseen situations or compensate for inaccurate or improper vehicle's decisions. For such cases, video feed, in addition to other data such as LIDAR, is considered essential to provide humans with the real picture of what is hap-pening to eventually take the right decision. However, if the video is not delivered in a timely fashion,it becomes useless or likely produce catastrophic outcomes. Additionally, any disruption in the streamed video, for instance during handover operation while traversing inter-countries cross borders, is very annoying to the user and possibly ause damages as well. In this article, we start by describing two important use cases, namely Remote Driving and Platooning, where the timely delivery of video is of extreme importance [1]. Thereafter, we detail our implemented solution to accommodate the aforementioned use cases for self-driving vehicles. Through extensive experiments in local and LTE networks, we show that our solution ensures a very low end-to-end latency. Also, we show that our solution keeps the video outage as low as possible during handover operation.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
