Nearly a hundred driverless cars stalled on Wuhan expressway — why Luobo Kuaipao (萝卜快跑) wasn’t ready

What happened

Nearly a hundred vehicles from Baidu (百度)’s autonomous taxi service Luobo Kuaipao (萝卜快跑) reportedly froze on Wuhan’s elevated ring road on the night of March 31, leaving multiple cars stopped in fast-moving lanes and trapping passengers until traffic police intervened. Wuhan’s traffic police issued a bulletin saying the episode was preliminarily attributed to a system fault and that no one was injured; investigators are still probing the root cause. It has been reported that passengers saw “system abnormal” on vehicle screens, could not reach customer service, and in some cases had to wait nearly two hours before being helped off high‑speed viaducts.

Technical and operational questions

Why did many cars stop at once? Experts and passengers pointed to a centralized failure mode: when cloud links or central servers hiccup, vehicles may execute a “minimal risk strategy” that brings them to a stop. That strategy makes sense in low‑speed, open environments. But on a high‑speed expressway, an in‑place stop can create greater danger than controlled relocation to the hard shoulder. It has been reported that early customer‑service responses blamed “network problems,” reinforcing the suspicion of large‑scale connectivity or cloud issues rather than isolated sensor faults. Similar episodes have happened abroad — Waymo in San Francisco once triggered a stop‑strategy when a traffic light failed — underscoring that the problem is systemic, not uniquely Chinese.

Scale, trust and regulatory implications

Luobo Kuaipao began full driverless operations in Wuhan in 2022 and, by 2024–25, had scaled to thousands of vehicles and millions of trips, part of China’s rapid push to commercialize autonomy. Scale raises stakes. When a centralized architecture fails, a single point of failure can become a city‑level traffic emergency. So who should have the control when the cloud goes dark — the operator, the passenger, or city authorities? Regulators may now press for non‑networked, physical “manual release” mechanisms, guaranteed seconds‑level interfaces between operators and traffic management, and stronger on‑street contingency obligations. Against a backdrop of U.S.–China tech tensions and shifting supply chains, China’s fast deployment of AV fleets makes clear that reliability and fail‑safe design are now as much a policy challenge as a technical one.

Company response and next steps

Baidu has not yet publicly disclosed a technical explanation or a passenger compensation plan. For Chinese and international observers alike, the Wuhan incident is a reminder: autonomous systems must plan for cascades, not only individual failures. Can the industry design architectures that avoid “all vehicles stop” outcomes? Regulators, operators and cities will be watching for answers — and for faster, clearer emergency controls that protect people rather than traffic metrics.