Why are self-driving taxis popular, and what is the experience in the United Sta

This is the "Future Trends" series from MIT Technology Review, spanning various industries, trends, and technologies, taking you to glimpse the future together. This is the "Future Trends" series from MIT Technology Review, spanning various industries, trends, and technologies, taking you to glimpse the future together. Recently, autonomous taxis have sparked widespread discussion in China, how is this issue viewed in the United States?

This article was published at the beginning of the year, with the original title:

What's next for robotaxis in 2024.

In 2023, the promising future of autonomous taxis seemed within reach. In San Francisco, summoning an autonomous taxi once became a trend, as simple and everyday as ordering food through an app. However, this dream shattered in October when a vehicle of Cruise, one of the leading autonomous taxi companies in the United States, was involved in a serious accident in downtown San Francisco, casting doubt on the future of the technology.

Another accident that followed led to an indefinite suspension of Cruise's services in California, and the National Highway Traffic Safety Administration also launched an investigation into the company. Since then, Cruise has withdrawn all vehicles and laid off 24% of its staff.Despite this, other autonomous taxi companies are still moving forward. In some cities in the United States and China, companies like Waymo and Baidu are still providing services to passengers willing to try their autonomous taxi fleets. Regulatory authorities in places like San Francisco, Phoenix, Beijing, Shanghai, and Wuhan now allow these vehicles to be driven without human safety operators.

Advertisement

However, other challenges still exist. Autonomous taxi companies need to get a return on their substantial investments. They cannot effectively compete with traditional taxis and Uber until the cost of autonomous taxis is reduced. If these companies expand their business too quickly, they may follow in the footsteps of Cruise. Another industry giant, Waymo, has adopted a more cautious strategy, but no company can completely avoid accidents.

Professor Missy Cummings, director of the Mason Autonomous and Robotics Center at George Mason University, said: "Once an accident occurs, it becomes big news, which will hurt the entire industry. This has been a big lesson this year, and the entire industry is walking on thin ice."

MIT Technology Review interviewed experts to explore the challenges faced by the autonomous taxi industry and how they expect the industry to change by 2024.

Funds, funds, and more fundsAfter years of road testing, various companies have demonstrated that the current version of autonomous driving technology is ready for deployment, albeit with some strict limitations. They operate only within strictly predefined geographical areas; although some vehicles no longer require a human operator in the driver's seat, a remote operator is still needed to take over in emergency situations; and they are limited to warm climate areas because snow can interfere with the vehicle's visual and sensory systems.

Associate Professor of Robotics and Mechanical Engineering at the University of Michigan, Ramanarayan Vasudevan, points out: "Based on publicly disclosed information, these systems still rely on a certain degree of remote human supervision to operate safely, so I call it automation rather than autonomy."

The problem is that the cost of this version of automated driving is much higher than that of traditional taxis. He said that the cost of autonomous taxis could be several orders of magnitude higher than the costs of other taxi companies. "Unfortunately, I believe that this technology will not undergo significant changes in the coming year to significantly reduce costs."

The high price will undoubtedly suppress demand. If autonomous taxis want to retain customers, they must not only attract curious first-timers but also make the service cheaper than other modes of transportation.

Bryant Walker Smith, Associate Professor of Law at the University of South Carolina, agrees: "These companies are competing with Uber drivers who earn less than the minimum wage standard, own a mid-range car, and maintain it themselves."In comparison, autonomous taxis are expensive vehicles filled with cameras, sensors, and sophisticated software systems, and they require continuous human monitoring and assistance. At least before a large number of autonomous taxis hit the road, they are almost unable to compete with ride-sharing services.

Smith also points out that as autonomous taxi companies continue to consume investors' funds, there is growing concern about the insufficient returns on their substantial expenditures. This means that there is an increasing urgency to generate performance while balancing potential benefits and costs.

Barriers to scaling up expansion

Currently in the United States, there are four cities where you can take an autonomous taxi: San Francisco, Phoenix, Los Angeles, and Las Vegas.The terms vary from city to city; some require joining a waiting list first, which may take months to get through, while others only operate vehicles within a small area.

Expanding autonomous taxi services to new cities requires significant upfront effort and cost: including detailed map-making (which also needs to be constantly updated) and adding new vehicles to cope with the growth in demand.

In addition, an autonomous driving system optimized for a specific city (such as San Francisco) has limited adaptability in other cities (such as Austin). Professor Cummings is researching how to assess this adaptability, "If I regard it as a fundamental research problem, it may mean that the company has not yet learned something important."

These factors together have sparked a new round of concerns about the profitability of autonomous taxis. Even though Cruise has withdrawn its vehicles, another major autonomous taxi company in the United States, Waymo, is not in a hurry to fill the gap. Because the cost of each autonomous taxi service is currently higher than the revenue, the motivation for large-scale expansion is insufficient.

Global Development SituationNot only is the United States researching, testing, and even deploying autonomous taxis.

Currently, China is another leader in this field, with a development timeline roughly synchronized with that of the United States. In 2023, several Chinese cities, including Beijing, Shanghai, Guangzhou, and Wuhan, were approved by the government to operate driverless taxis without safety operators. However, these vehicles can only travel in some smaller and relatively remote areas of the city, making the service difficult to access for most people.

With the support of Chinese and American companies, the Middle East is also rapidly rising in this field. Saudi Arabia invested $100 million in the Chinese autonomous taxi startup Pony.AI, introducing its vehicles to the future city Neom, which is under construction and is said to be built with all the latest technologies. At the same time, Dubai and Abu Dhabi are competing to become the first city in the Middle East to pilot autonomous vehicles on the road, manufactured by Cruise and Chinese company WeRide.

Chinese autonomous taxi companies face the same core challenges as their American counterparts: proving their profitability. In 2023, a commercialization trend emerged in the Chinese industry, with Chinese autonomous driving companies competing to sell their autonomous driving systems to other companies, quickly realizing profits by repackaging the technology into services with higher demand but slightly lower technical content (such as urban autonomous driving systems, which can be sold to car manufacturers).

In contrast, Europe's development pace in the field of autonomous taxis is slower, partly because European countries are more inclined to deploy autonomous driving technology in public transportation. Although there are road tests for autonomous taxis in Germany, the UK, and France, commercial operations are still far from reality.Lessons from the Cruise Incident

The painful experience of Cruise highlights one of the main obstacles faced by autonomous taxi services: they sometimes exhibit abnormal behavior. In October 2023, in San Francisco, a human driver (driving a non-autonomous vehicle) hit a pedestrian and fled the scene. Subsequently, a passing Cruise vehicle ran over the victim and stopped after dragging them for 20 feet.

Cathy Chase, President of Advocates for Highway and Auto Safety based in Washington, D.C., said, "We are deeply concerned that more lives will be lost, more first responders will be obstructed, and more sudden stops will occur. We are not against autonomous vehicles; what we worry about is the unsafe deployment and the market's eagerness to sacrifice public travel safety for quick success."

She said that the data provided by these companies is insufficient to demonstrate the safety of their vehicles. Although they must submit data to the National Highway Traffic Safety Administration, the data is often heavily redacted before publication to protect trade secrets. Chase said that some federal bills proposed in the past year, which did not pass, could even further reduce these reporting requirements.Cummings believes that this accident will force everyone to face the complexity and uncertainty of the operation of autonomous taxis, and may prompt the industry to pay more attention to remote human supervision, which is what the Cruise vehicles may have lacked in that accident. However, increasing human intervention will undoubtedly exacerbate the profitability problem.

At the same time, the California Public Utilities Commission accused Cruise of misleading the public and regulators on the issue of responsibility for the accident. "If we can't trust these companies, then they have no business operating on our roads," Smith said.

A spokesperson for Cruise told MIT Technology Review that the company currently has no new information to share, but mentioned a blog post in November 2023, which revealed that a third-party law firm and technology consulting company have been hired to review the accident and the response to regulation. In the settlement proposal submitted to the CPUC (California Public Utilities Commission), Cruise also proposed to share more data, including "collision reports and regular reports on incidents involving the cessation of autonomous driving vehicles."

The future of Cruise is still unclear, and the company's original plan to start operations in several cities soon has also been suspended. Meanwhile, Waymo is applying to expand its services in Los Angeles and introduce its vehicles to the highways of Phoenix. And Amazon's autonomous driving startup Zoox seems to be preparing to launch services in several unspecified cities, which shows that despite facing challenges, other participants in the autonomous taxi industry are still continuing to explore and advance.