美國終止特斯拉智慧召喚安全調查

🔑 增強重點摘要

• •The National Highway Traffic Safety Administration (NHTSA) investigation specifically focused on the 'Actually Smart Summon' (ASS) update, which utilizes Tesla's end-to-end neural network architecture rather than the legacy Summon code.
• •Tesla implemented several over-the-air (OTA) software updates during the pendency of the probe that improved object detection and path planning, which contributed to the agency's decision to close the case.
• •The closure of this probe does not preclude future regulatory action if new data suggests a pattern of safety issues, as the NHTSA maintains ongoing monitoring of all Advanced Driver Assistance Systems (ADAS).

🛠️ 技術深入

• •Transitioned from heuristic-based code to a vision-based, end-to-end neural network (v12 architecture) that processes raw video input to output steering, braking, and acceleration commands.
• •Utilizes Occupancy Networks to create a real-time 3D representation of the environment, allowing the vehicle to identify and navigate around dynamic obstacles like shopping carts or pedestrians.
• •Incorporates 'Vector Space' mapping, which allows the vehicle to understand lane markings, curbs, and parking spot boundaries without relying on pre-existing high-definition maps.

🔮 前景展望基於引用來源的 AI 分析

⏳ 時間線