A recent incident involving Baidu's autonomous ride-hailing service, Baidu Robotaxi, has highlighted vulnerabilities in the communication networks supporting self-driving vehicles. On the evening of March 31, 2026, at 8:57 PM, a Baidu Robotaxi vehicle slowed down abruptly on Wuhan's Third Ring Road before coming to a complete stop in the fast lane. Within ten minutes, nearly 100 white autonomous vehicles experienced similar failures simultaneously on major routes including Taizihu Bridge, Baishazhou Bridge, the Second Ring Road, and Yangsigang Bridge.
Passengers reported being stranded on elevated roads for nearly two hours, with in-vehicle SOS buttons unresponsive and customer service lines engaged. Traffic police eventually conducted a rescue operation on foot. Wuhan traffic authorities later issued a preliminary statement attributing the incident to a system malfunction.
This was not the first occurrence of such failures for Baidu Robotaxi. In July 2024, vehicles similarly stalled during Wuhan's evening rush hour, requiring police intervention after contacting customer service to move the cars to roadside locations. The recurrence of identical "system malfunctions" and "police interventions" after nearly two years raises concerns.
Baidu Robotaxi represents Baidu's core business in intelligent driving and mobility services, serving as a crucial commercialization channel for Baidu's AI technologies. According to Baidu's 2025 annual report, the operational entity is Robo Force (Beijing) Technology Co., Ltd., fully owned by Baidu's Apollo Intelligent Technology (Beijing) Co., Ltd. Neither Baidu nor Robo Force has publicly disclosed the cause of the recent malfunction.
Industry analysis from Guidehouse Insights' Q4 2025 report identifies Baidu Apollo as one of two global leaders in autonomous driving, alongside U.S.-based Waymo. Baidu Robotaxi's technology relies on Baidu's four-layer AI architecture encompassing cloud infrastructure, deep learning frameworks, large models, and applications.
As of February 2026, Baidu Robotaxi has expanded to 26 cities globally. Since February 2025, it has achieved 100% unmanned operations (L4 level) across all Chinese cities including Beijing, Shanghai, Shenzhen, Wuhan, Chengdu, Chongqing, Haikou, and Sanya, with approved fee structures. Cumulative rides provided to the public exceed 20 million. Internationally, services operate in Dubai, Abu Dhabi, London, St. Gallen, and Seoul. Notably, while Baidu Robotaxi launched fully commercial unmanned operations in Dubai on March 30, the mass stall incident occurred in Wuhan the following day.
Professor Zhu Xichan from Tongji University's Automotive and Energy College, a leading expert in China's intelligent connected vehicle research, provided analysis. Professor Zhu, who helped establish China's first automotive collision safety standards, currently focuses on intelligent driving testing and standardization.
Professor Zhu clarified that current Robotaxi operations rely on "connected intelligence" rather than purely autonomous vehicle intelligence. The system integrates five components: vehicle, road, cloud, network, and mapping. The March 31 failure likely originated in the network component. Contrary to common assumptions, Robotaxi operations currently use consumer-grade cellular networks rather than specialized 5G infrastructure, making network instability and disconnections unavoidable.
Regarding vehicles stopping in travel lanes rather than pulling over, Professor Zhu acknowledged this as a safety measure but criticized the implementation. While L4-capable vehicles possess high-definition maps and autonomous lane-changing capabilities, the system defaulted to immediate stopping rather than navigating to safe locations. He suggested improvements to minimum risk strategies, including programmed responses to guide vehicles to roadside safety during network failures.
Professor Zhu addressed response failures, noting that cloud-based supervision platforms become ineffective during network outages, requiring ground personnel intervention. Congestion from stalled vehicles further delayed response times. He compared the incident to a Waymo shutdown in San Francisco four months earlier during a city-wide power outage.
On liability, Professor Zhu indicated that current regulations fall under Wuhan's Intelligent Connected Vehicle Demonstration Zone guidelines rather than traditional automotive laws. Regarding expansion pace versus public safety, he emphasized that demonstration operations aim to identify problems for rapid iteration, balancing technological progress with safety requirements.
Professor Zhu offered three recommendations: improved public communication strategies during incidents, optimized minimum risk protocols for network failures, and continued development of reliable communication solutions for vehicle-cloud connectivity. When asked about personal usage recommendations, Professor Zhu responded that while daily commuting is acceptable, users should maintain awareness of the technology's current limitations.
Comments