特斯拉上线 Autopilot

2015 年 10 月，特斯拉 Model S 车主收到一次软件更新，车辆突然多了个叫 Autopilot 的功能：车道保持、自动变道、自适应巡航。没有新车发布，没有换代，一辆已经交付的车，靠一段代码升级就获得了新的驾驶能力。这在当时的汽车行业几乎不可想象。

在 2015 年的自动驾驶叙事里，主流想象是「未来某天会有无人出租车队」，Google 的测试车是那个方向的代表。特斯拉选了另一条路：先让辅助驾驶大规模上路，让真实的驾驶员和真实的道路替自己收集数据。这条路更慢、更受争议，但它有一个 Google 测试车给不了的东西——海量真实驾驶里程。

Autopilot 上线后，争议几乎同步到来。它被定位为需要驾驶员时刻留神的辅助驾驶，但名字本身就在暗示「自动驾驶」，事故随之引发公众对辅助驾驶边界的大讨论。即便如此，它确实在技术上跑通了一个循环：车上传感器收集数据，云端模型迭代，OTA 推送新版本，车越开越「聪明」。

这个模式对整个行业的影响远超特斯拉本身。传统车企开始正视「软件定义汽车」，辅助驾驶成了购车决策里的核心卖点；端到端驾驶数据与 OTA 迭代成为主流思路。可以说，2015 年那一次 OTA 推送，把整个汽车行业的研发重心推向了 AI 与数据。

回看 Autopilot 的上线，它最了不起的地方是重新定义了「自动驾驶」进入世界的方式：不是等完美，而是先上路、再进化。今天无论是 Waymo 的无人出租车，还是国产车企的智驾竞赛，背后都有一根线连着 2015 年那个秋天的软件更新。

In October 2015 Tesla Model S owners received a software update, and their cars suddenly gained a feature called Autopilot: lane keeping, automatic lane changes, adaptive cruise. No new model, no redesign—an already-delivered car gained new driving ability through a code update. In the auto industry of 2015 this was almost unthinkable.

Back then the mainstream autonomy narrative was "one day there will be robotaxi fleets", with Google's test cars as the standard bearer. Tesla chose another path: ship driver assistance at scale first, let real drivers and real roads collect the data. Slower and far more controversial, but it produced something Google test cars never had—massive real-world driving miles.

Controversy arrived almost immediately. Autopilot was positioned as driver assistance requiring constant attention, but the name itself whispered "self-driving". Accidents followed, and the public debate over the boundaries of driver assistance began. Technically, though, it closed a loop: on-board sensors collect data, cloud models iterate, OTA pushes new versions, and the car grows smarter with every mile.

The impact reached far beyond Tesla. Traditional automakers faced "software-defined vehicles"; driver assistance became a core selling point; end-to-end driving data and OTA iteration became the standard thinking. That one OTA push in 2015 redirected the entire industry's R&D toward AI and data.

Looking back, Autopilot's greatest achievement was redefining how autonomy enters the world: not wait for perfection, but ship, then evolve. Whether Waymo's robotaxis today or the smart-driving race of Chinese automakers, a thread runs back to that software update in the fall of 2015.