Baidu’s robotaxis froze in traffic creating chaos
Foto: Bloomberg via Getty Images
More than 100 Apollo Go autonomous taxis belonging to the Chinese giant Baidu paralyzed traffic in Wuhan, turning city arteries into a zone of communication chaos. On Tuesday, a massive system failure occurred, resulting in robotaxis stopping without warning in the middle of roads and highways, trapping passengers inside the cabins. Local authorities confirmed numerous reports of congestion and at least one collision caused by the sudden "freezing" of the fleet. Although preliminary findings point to a system error, the incident casts a shadow over the safety of autonomous technologies at a time of their aggressive expansion. For users worldwide, this is a warning signal regarding the stability of AI infrastructure in public transport. Baidu currently operates in 26 cities globally and collaborates with Uber in metropolises such as London and Dubai, meaning similar glitches could become an international problem. The events in Wuhan demonstrate that despite advanced algorithms, the absence of a physical driver in critical situations still generates risks that modern engineering cannot fully eliminate. The scale of this paralysis forces a renewed debate on the necessity of implementing remote emergency systems and more rigorous safety protocols in driverless vehicles.
The visualization of an autonomous future, where fleets of driverless vehicles move seamlessly through urban arteries, has just suffered a major blow to its image. On Tuesday, April 1, 2026, the streets of Wuhan became the scene of unprecedented traffic paralysis. Hundreds of passengers using the Apollo Go platform, owned by the tech giant Baidu, were stranded in vehicles that suddenly and without warning refused to cooperate. The failure was not limited to side streets – robotaxis "froze" on highways and major intersections, leading to chaos on a scale that forced intervention by law enforcement services.
This event is particularly significant because Wuhan serves as a global testing ground for Baidu. The company operates a fleet of over 500 autonomous cars there, building one of the most advanced driverless transport systems in the world. The sudden immobilization of a significant portion of this fleet – estimated to have affected at least 100 vehicles – exposes critical weaknesses in systems based on artificial intelligence when they are thrown into a real, dense urban environment. Wuhan police confirmed numerous reports of road blockades, and preliminary findings point to an "unspecified system failure" that paralyzed the decision-making process of the control algorithms.
When the algorithm says "stop" at the worst moment
The problem with the Apollo Go robotaxis was not just that they stopped, but where it happened. Reports indicate that vehicles immobilized in the middle of traffic lanes, on highway exits, and in tunnels, creating traps for the passengers traveling in them. Some were trapped inside the cars for an extended period, unable to force the system to continue driving or to safely exit the vehicle in high-traffic conditions. This situation led to at least one road accident, although fortunately, police reported no fatalities or serious injuries.
Read also
From an engineering perspective, the "freezing" of the fleet suggests an error in the communication layer or the central decision module (a so-called system failure). In Level 4 autonomy, which Baidu promotes, a vehicle should have emergency procedures allowing it to safely pull over to the shoulder if a fault is detected. The fact that dozens of cars simply stopped in place, blocking traffic, indicates a total loss of operational capability in the Apollo software. For the Autonomous Cars industry, this is a warning sign: even thousands of kilometers driven without incident do not guarantee immunity to a cascading software failure.
It is worth noting Baidu's operational context. The company is not limited to the local market. It currently manages robotaxi services in 26 cities around the world, including key metropolises like London and Dubai, where it collaborates with Uber. The incident in Wuhan calls into question the safety of the global expansion of these systems. If a system can fail so spectacularly in a city that is a "home" market for Baidu, regulatory bodies in Europe or the Middle East may revise their approach to issuing licenses for driverless transport.
Scale of the problem and urban infrastructure response
Tuesday's events show that urban infrastructure is not yet prepared for mass failures of autonomous vehicles. When a traditional car breaks down, the driver can attempt to push it or call for help. In the case of Apollo Go, locked systems and the lack of a physical driver meant that law enforcement in Wuhan had great difficulty clearing the routes. The following list illustrates key aspects of this failure:
- Scale: Over 100 immobilized vehicles at key points in the city.
- Location: Wuhan, Baidu's main technological hub.
- Cause: General system failure preventing movement.
- Consequences: Traffic paralysis, trapped passengers, road collisions.
- Market reach: Baidu operates in 26 cities, giving the incident global significance.
For tech market observers, such as Robert Hart from The Verge, this situation is proof that AI technology in transport is still in a high-risk phase. Although Baidu has not yet published a detailed technical report regarding the causes of the outage, pressure from the public and local authorities is mounting. Wuhan has so far been an example of the successful implementation of Apollo Go, but Tuesday's chaos could drastically change residents' attitudes toward sharing the road with robots.
Analyzing this case, it should be noted that the problem is not autonomous driving technology itself, but its resilience to critical errors. In a world where Baidu aims to completely replace taxi drivers with algorithms, the lack of "fail-safe" mechanisms is unacceptable. This incident will certainly be used by opponents of rapid autonomy adoption, pointing out that human oversight is still necessary, at least as an ultimate safety fuse in emergency situations.
Crisis of confidence in autonomous mobility
The Tuesday failure in Wuhan is more than just a technical glitch; it is a serious blow to social trust. Baidu, in promoting Apollo Go, emphasized safety and the elimination of human error. Meanwhile, machine error proved to be just as, if if not more, paralyzing. Passengers trapped on highways experienced a fear that cannot be easily erased by marketing campaigns about the bright future of transport.
In light of Baidu's global ambitions and their partnerships with giants like Uber, it will be necessary to implement new standards of transparency. The industry must answer the question: what happens to a passenger when the "brain" of the car stops working in a high-traffic zone? Without clear answers and guarantees that the situation in Wuhan will not happen again, the development of robotaxis could be slowed down by rigorous regulations appearing as a reaction to this chaos.
One could argue that the incident in Wuhan will become a turning point in the regulation of AI systems in public transport. This experience will force manufacturers like Baidu to move away from focusing solely on navigational capabilities and instead focus on building redundant, physical safety systems that allow for remote or manual extraction of the vehicle from a critical situation. Without this, the vision of cities full of robotaxis will remain only a risky experiment that could turn into a traffic nightmare at any moment.
