HUBEI, Central China — At a busy bus stop in central China’s city of Wuhan, the roads come alive during rush hour with a curious mix of old and new. Traditional taxis, with their glowing “occupied” signs, now share lanes with sleek, driverless robotaxis, each topped with a spinning laser radar system.

As both weave through traffic, honking furiously at each other, residents have taken notice — playfully dubbing the newcomers shaoluobo, or “simple-minded radish.” The name is a portmanteau of Baidu’s robotaxi service, Luobo Kuaipao (Apollo Go in English), and slang from the local dialect.

Despite the playful nickname, these self-driving taxis are gradually finding their place in Wuhan’s transportation system. This integration is part of a massive push by Chinese authorities and tech giants to position autonomous vehicles at the forefront of urban innovation.

But this rapid rollout has not been without controversy. While some see these robotaxis as a sign of progress — promising safer roads, cheaper rides, and reduced emissions — others worry about the disruptions they bring. Concerns range from affecting traditional taxi drivers to the readiness of city infrastructure to handle such advanced technology.

Yet, robotaxis are here to stay. With over 400 in operation and plans to reach 1,000 by the end of 2024, Baidu is betting on this service, expecting profitability in Wuhan by 2025. The popularity of these self-driving taxis, fueled by Baidu’s cheap fares, has the company eyeing expansion into other Chinese megacities, supported by eager local governments.

On Wuhan’s streets, it’s now common to see passengers waiting for their rides, and in coffee shops, customers can often be heard swapping stories about their experiences — good and bad — with the automated vehicles.

Over three days, Sixth Tone joined them, riding the robotaxis through the city, from morning rush hour to late at night, witnessing firsthand how this cutting-edge technology is reshaping life in Wuhan.

Hopping in

Ordering a robotaxi in Wuhan isn’t much different from hailing a traditional ride-hailing service. Passengers set their pickup and drop-off spots via an exclusive mobile app, just as they would with any other service.

But when the sleek, driverless vehicle arrives, it’s far from typical.

To unlock the car, passengers must input the last four digits of their phone number into a touchscreen on the door. Once inside, the car remains stationary until everyone is buckled up and the “start” button on a display inside is pressed — only then does the journey begin.

Inside, there are more noticeable differences. A glass barrier separates the rear seats from the front, along with a sign that reads, “Please do not enter the front area or touch any equipment.” A small mirror in the back allows passengers to check the road conditions, and the touchscreen display lets them choose their preferred music for the ride.

Though the front area is off-limits, it contains driver and passenger seats, either of which may be used by an on-site safety operator or left empty while a remote operator supervises the journey. The safety operator intervenes only when the robotaxi encounters an issue.

Each car also includes a one-page manual, written entirely in Chinese, that outlines the dos and don’ts of the service. It emphasizes basic rules, such as staying buckled up at all times, and informs passengers that both voices and images will be recorded during the ride — a detail that might give some riders pause.

In Wuhan’s blistering summer heat, the robotaxi’s air conditioning provides a cool refuge, and with no human driver on board, there’s no worry about lingering scents or unwanted small talk.

But the service isn’t without its drawbacks. The robotaxi’s trunk is unavailable for use, and with the front seat off-limits, the vehicle can only accommodate up to three passengers. Large luggage is also a no-go, which can be frustrating for those traveling with more than just a few bags.

While these inconveniences might be manageable for some, many passengers emphasize that the bigger issue is availability. Despite recent expansions — including more than 3,379 kilometers of one-way roads dedicated to self-driving trials, the most in the world — robotaxi services are still not accessible in the most central districts of Wuhan, leaving many potential passengers without options.

For now, densely populated areas within operating districts are not covered. For instance, the popular Guanggu pedestrian street, home to a large number of shopping malls, is not open for the robotaxi rides, and the nearest drop-off spot is kilometers away.

Technical difficulties

Major Chinese cities have been testing self-driving vehicles for years, with authorities and developers hoping they will ease traffic, improve safety, and reduce emissions. So far, 21 cities have launched autonomous driving pilots, with 60 companies obtaining testing permits.

But Wuhan is at the heart of this autonomous driving push. As one of the first cities to embrace the technology, it now boasts the world’s largest testing area for self-driving cars, nearly 25 times the size of San Francisco. Unlike other cities where trials are confined to suburban areas, Wuhan has expanded testing into its busy downtown.

In May, Wang Yunpeng, vice president of Baidu’s autonomous driving division, emphasized the advancements of Apollo Go’s fleets in Wuhan, particularly in comparison to Tesla’s robotaxi plans.

“In Wuhan, it will take at least three years, or potentially up to five years, for Tesla to reach the current level of fully driverless operation that Baidu has achieved,” Wang stated. “They can try the service if they don’t believe it.”

While the city and Baidu have made strides, the technology is still far from perfect.

Public concerns about the robotaxi revolution came to the forefront in July after an Apollo Go vehicle in Wuhan collided with a pedestrian. Footage of the incident quickly spread across social media, igniting a heated debate about the safety and broader implications of autonomous vehicles.

Passengers also report that robotaxis struggle during persistent rain, often becoming unavailable in such conditions. On social media, some claimed that the self-driven car stopped for an hour amid a downpour and started running only after the rain stopped.

The complexity of Wuhan’s transportation system adds to the difficulties, with current self-driving vehicles struggling to navigate mixed traffic flows that include cars, bicycles, and pedestrians.

According to a report from Beijing’s Tsinghua University, autonomous vehicles still fall short in scenarios involving sudden obstacles or adverse weather conditions like rain, snow, and fog. The sensors, essential for detecting objects and the environment, often underperform in these elements, making safe navigation difficult.

Local residents and drivers also point out another critical issue: the robotaxis operate at relatively low speeds — usually below 50 km/h, with a maximum of 70 km/h — and struggle in heavy traffic.

This often leads to increased congestion, as the vehicles are easily overwhelmed, causing slow responses and unexpected stops that disrupt traffic — bad news in a city that already ranks as the fifth most congested in China during commuting hours, according to Baidu Maps.

In recent months, a surge of complaints about robotaxis has emerged on a government website, with many criticizing the vehicles for occupying fast lanes while moving at slow speeds. Some have called on authorities to limit or even cancel robotaxi operations during rush hours.

While Baidu earlier denied allegations that their robotaxis are causing traffic accidents, dismissing some of the complaints as “malicious,” the company did not respond to Sixth Tone’s request for comment. However, a member of staff from Baidu’s customer service explained that the vehicles only stop when they detect an obstacle in front of them and that such stops are in full accordance with traffic rules.

The robotaxis’ overemphasis on safety was evident during Sixth Tone’s rides. In one instance, a driverless car crawled through a tunnel at under 30 km/h, though it was the only vehicle on the road. In another, instead of overtaking a bicycle, the robotaxi slowed down to follow it instead.

According to Deng Zhidong, a professor at Tsinghua University specializing in autonomous driving, the issue largely stems from the robotaxis’ conservative driving strategy, which aims to minimize the risk of accidents.

“Traffic downtown is dense, and driving is relatively more complex with shorter safety distances. This requires more reliable single-vehicle intelligence and more agile remote safety takeover technology,” says Deng, adding that developers have consciously adopted more conservative driving strategies for self-driving vehicles.

“A more aggressive strategy might appear more human-like, but it’s also more likely to cause accidents,” he asserts. “Machines don’t get distracted, they don’t drive angry, and they never operate a vehicle while intoxicated or fatigued.”

Deng adds that with the rapid advancement of AI technologies, many self-driving enterprises are now exploring a new end-to-end technical paradigm. This approach aims to teach autonomous systems how to learn and acquire driving skills, enabling them to navigate different scenarios by observing, processing information, and making dynamic decisions, rather than just following pre-programmed rules.

Shu Zhan, a ride-share driver in Wuhan who has witnessed the rapid expansion of self-driving cars firsthand, is proud of the city’s progress but admits the robotaxis are causing significant disruptions. “Since such mostly-minor issues are now solved by human intervention, you can’t just leave the robotaxis unattended as the service expands to larger areas.”

The road ahead

Beyond the technological challenges, Shu and other traditional taxi drivers in Wuhan tell Sixth Tone that they now stare at an uncertain future. As robotaxis expand into more areas, many worry about the new kind of competition these machines represent — vehicles that don’t need rest and can operate around the clock.

Many also admit that they’re keeping their distance from robotaxis for now, particularly with increasing social media posts of tense encounters between robotaxis and other vehicles. “It’s not about the need to drive faster,” explains Shu, “but about having no one to address when a collision occurs and no clear way to handle the situation.”

Current road traffic safety laws, which focus on the responsibilities and qualifications of human drivers, don’t adequately address situations where a vehicle is fully controlled by an AI system, according to Yu Lingyun, a law professor at Tsinghua University.

Despite several guidelines, China has yet to establish a comprehensive law specifically addressing autonomous driving. The national guideline for conducting self-driving tests states that in the event of a traffic incident, “the responsibilities of the parties should be determined in accordance with the national road traffic safety law and other regulations.”

Some regional regulations have provided further clarification. For instance, a guideline in Wuhan stipulates that the owner or administrator of an autonomous vehicle should be held accountable in the event of an accident. They are also granted the right to seek compensation from other parties involved.

Zhu Wei, a law professor at China University of Political Science and Law, called for clearer laws and regulations to address liability issues as self-driving vehicles involve multiple parties, from drivers to software and hardware providers. He also stresses the need for more comprehensive insurance policies tailored to autonomous driving, with specific rules on compensation.

“The future challenges of the autonomous driving industry aren’t primarily about technical glitches but about non-technical factors such as traffic accident liability, insurance, and compensation,” asserts Yu. “These issues can’t be resolved through technological advancements alone; they require legislative regulation.”

Despite these challenges — whether technical, legal, or social — some drivers remain unconcerned.

“Machines are machines; they don’t have that kind of warmth,” says Wu, a veteran ride-share driver who requested to be identified only by his surname. Wu believes human drivers still offer something irreplaceable: the ability to adapt to the nuances of the road and provide personalized service. “We can adjust our driving based on the situation and the needs of our passengers, something a robotaxi just can’t do,” he says.

Still, Wu admits that driverless cars are likely to become part of the system as they become more competitive thanks to lower operating costs. He predicts that as the city’s congestion eases and traffic conditions improve, the environment will be more conducive to the operation of these vehicles.

Asked if the potential for widespread adoption concerns him, the veteran driver was unfazed. “Concerns and complaints mean nothing,” says Wu. “Autonomous driving represents technological and social progress; it cannot be stopped. The only thing we can do is adapt.”

Editor: Apurva.

(Header image: Inside a a driverless robotaxi in Wuhan, Hubei province, 2024. VCG)