news

The large-scale commercial travel service provided by Luobo Kuaipao in Wuhan has recently sparked heated discussions. The biggest reaction came from local taxi drivers, whose order volume was directly impacted by driverless cars.

Also impacted by the AI ​​technology developed by major companies, professions such as human painters and screenwriters have received widespread sympathy and support. In travel scenarios, a large number of netizens have stood on the side of smart machines and supported Luobo Kuaipao in "rectifying" human driver behaviors such as illegal driving and poor service.

Some people even put forward conspiracy theories: It is really smart to deploy driverless taxis in Wuhan. Other places may not be very confident about AI, but in Wuhan, it will not be worse.

What is so special about Wuhan?

Wuhan drivers have long been known on the Chinese Internet for their wild driving, and Wuhan is therefore considered "the city that does the best job of pedestrians giving way to motor vehicles." In this case, it can be said that cars (drivers) enjoy the greatest right of way.

The conflict over "road rights" between people and vehicles in Wuhan is very prominent.

The Carrot Fast Run driverless car that obeys traffic rules can not only give way to pedestrians and protect the right of ordinary people to travel, but also create "trouble" for illegal vehicles, and make the other party bear most or even all the responsibility when friction occurs, so as to "vent" the anger of pedestrians. Naturally, it has won the support of a large number of people.

An expert once told me: The end point of intelligent transportation isAutopilot. At present, 31 provinces and cities have mentioned the construction of major transportation infrastructure such as smart highways, autonomous driving (vehicle-road collaboration), intelligent transportation, and intelligent traffic management in their 2024 government work reports.

Obviously, the development of intelligent transportation is the general trend, and the purpose is not to make things difficult for taxi drivers, but to ultimately resolve the conflict over road rights between people and cars through technological progress.

There is reason to believe that a world with high traffic intelligence and harmonious human-vehicle relationship will be aSmart CityThe commercial and economic formats it carries are sufficient to absorb more people into better employment and life.

Let’s shift our focus to the present and talk about what hidden pains of today’s cities are exposed behind the conflicts between “overconfident drivers” and “disenfranchised passersby”.

First of all, I need to say something for Wuhan. Although the "super-strong driver" meme has become popular on the Internet, it does not mean that Wuhan drivers are particularly unruly. In fact, drivers (motor vehicles) fighting for the right of way with pedestrians is a common occurrence in cities across the country.

Everyone must have had this experience:

When crossing the road, you have to run fast and in small steps to barely pass the traffic light.

Urban roads are getting wider and wider, but it is the motor vehicle lanes that are being widened, while pedestrians and non-motor vehicles are still crowded on a small road.

Take Beijing, where I live, for example. The walking environment is not friendly to pedestrians. In this regard, Mao Qizhi, deputy dean of the School of Architecture at Tsinghua University, once publicly stated that because many people drive, fewer and fewer people choose to walk or other slow transportation, and the pedestrian paths are becoming increasingly "weak".

We all know that urban transportation should be "people-oriented". However, due to the rapid expansion of urban scale and the surge in the number of vehicles in megacities, large and medium-sized cities, the existing road capacity is saturated. Therefore, in order to ensure the efficiency of automobile traffic, priority is given to widening motor vehicle lanes, sacrificing the passage space for pedestrians and non-motor vehicles; adjusting the timing of traffic lights to shorten the time for pedestrians to cross the street; in the race against time, some motor vehicle drivers do not give way to pedestrians and arbitrarily cut in and change lanes, making pedestrians and civilized drivers miserable.

"In Wuhan, only sweet radish gives way to pedestrians." It is in this context that Radish Run becomes a messenger of justice and judge of the road.

Pedestrians are "weaker" than motor vehicles, and civilized drivers are "weaker" than illegal drivers. The "lose-lose situation" of pedestrians and drivers is ultimately caused by the contradiction between the surge in the number of cars and the efficiency of urban road travel.

In order to resolve the conflict between cars and roads, cities have come up with many solutions.

For example, vehicle restrictions and purchase restrictions can control the growth of the number of motor vehicles in cities, but this can only reduce the pressure on road carrying capacity and cannot solve the problem of traffic efficiency of existing vehicles.

Another example is that cars must give way to pedestrians, and those who violate the rules will be fined, in order to protect the basic right of pedestrians and non-motor vehicles to pass. Many urban drivers have been "fined out of character". However, when cars give way to pedestrians, drivers bear the cost of reduced efficiency, and are prone to rushing to save time by cutting in, cutting in, and driving fast, which in turn increases the accident rate and congestion risk.

Of course, there is another means, which is the introduction of new technologies.

In recent years, the number one scenario for the implementation of "smart cities" is intelligent transportation. Through a large number of sensors, smart cameras, and smart connected vehicles as the "sensory nerves" of intelligent transportation, the cloud-based algorithms and solutions of the "city brain" are used to optimize traffic light timing, rationally utilize tidal lanes, and provide early warning of congestion, thereby alleviating urban traffic pressure.

The construction of "smart cities" has indeed greatly reduced traffic jams, but there are two problems to be solved.

First, under the impact of the absolute number of motor vehicles, the optimization effect brought by the algorithm is limited.Beijing is still congested during rush hours in the morning and evening.Second, each region has its own intelligent transportation system. In the previous stage, "smart cities" were mostly promoted on a regional and departmental basis.For example, we have interviewed a district in Haidian, Beijing, which can monitor the routes of illegal dump trucks and take real-time photos of illegal motor vehicle behaviors, but it is not interconnected with other districts. Once a vehicle leaves the area, it cannot be monitored. This kind of data silo is common in the same city, which means that the algorithm can only work in a small area.

Autonomous driving technology is expected to fundamentally resolve the conflict between vehicles and roads in urban traffic and return the "right of way" to pedestrians.

It mainly plays a role in three aspects:

1. Public transportation.

Since the road traffic efficiency is low and the load-bearing pressure is high, then reducing the total number of motor vehicles can "cure the root cause". And self-driving cars can serve as "public drivers". Now some netizens are imagining: buy a "carrot" to pick you up and drop you off at work, and drive an online car-hailing service at other times. Therefore, passengers not only do not need to take human taxis, but also do not need to buy cars. Self-driving can directly reduce the demand for passenger cars, make more full use of road traffic space and driving time, and greatly improve urban travel efficiency.

2. Comprehensive intelligence.

As mentioned earlier, smart transportation could only function within a region before, and data could not be interconnected, which limited the algorithm's ability to optimize urban transportation.The large-scale commercial use of autonomous driving means that driverless cars must move from regional pilot projects to the entire city, which requires a unified, collaborative, and high-level intelligent transportation infrastructure.

It can be said that the large-scale commercial use of self-driving cars will directly accelerate the upgrading of digital and intelligent infrastructure in cities. For example, in order to support the safe passage of unmanned vehicles, wireless transmission needs to be upgraded from gigabit to5GThe upgrade to 10G 5.5G, extensive coverage of roadside sensing devices such as high-definition smart cameras/smart lamp poles, and a cloud platform "brain" that gathers real-time data of unmanned vehicles for unified scheduling have exponentially improved the decision-making capabilities of urban intelligent transportation.

3. Standardized driving.

"In the future, the roads may be full of driverless cars, and human driving will become a luxury item in special venues, just like equestrianism and racing." Once upon a time, a practitioner predicted this. When Luobo Kuaipao was operating in Wuhan, as a "road judge", once a collision or scratch occurred, it was basically the responsibility of the human driver for changing lanes illegally, driving aggressively, and not abiding by traffic rules.

Although the actual driving speed of driverless cars is still relatively slow today (which is also the biggest complaint), as the proportion of driverless cars on the road gradually increases, the perception, transmission and decision-making of data between cars can logically increase the driving speed while ensuring a safe distance and complying with traffic rules. In this way, the traffic efficiency is improved, the accident rate of driverless cars and human drivers is reduced, the personal safety of passengers is guaranteed, and the congestion caused by accidents is reduced.

To solve the problem of urban traffic pressure, we sometimes need to think outside the box. In addition to building more and wider roads and installing faster traffic lights, we can also use fewer cars to transport more people.

Of course, some people say that autonomous driving is a great idea in the long run, but taking away the jobs of taxi/online ride-hailing drivers now will bring pain to specific people.

Will this concern arise?

We mentioned at the beginning thatThe end point of intelligent transportation is autonomous driving。

In other words, fully autonomous driving is the ultimate form of future transportation. Worrying today that autonomous driving will destroy the travel market is like worrying about alternating current destroying cities in 1890 and worrying about AGI destroying humanity in 2024. This is "overestimating the short-term impact of technology."

On the one hand, the number of self-driving taxis currently in operation is very small among commercially operated vehicles.For example, at present, Luobo Kuaipao has only deployed more than 400 vehicles in Wuhan, while the number of registered cruising taxis in Wuhan is about 20,000, and the number of online car-hailing drivers is about 24,000 (according to a report in 2023), and driverless cars account for less than 1%.

In addition, the autonomous driving services available to the public are all at the "universal access in exchange for data" stage and do not have the sustainability for large-scale commercial use.

Luobo Kuaipao is currently operating at a small profit with subsidies in Wuhan. Shanghai has just issued the first batch of "completely unmanned vehicle licenses" at the 2024 World Artificial Intelligence Conference. Citizens can ride unmanned vehicles for free on some sections of Pudong. In June this year, Beijing just issued licenses to Pony.ai and WeRide to carry out commercial pilot projects of autonomous driving travel services between Beijing Yizhuang and the terminal of Beijing Daxing International Airport. Daxing Airport's autonomous driving shuttle does not require additional highway tolls.

It is not difficult to see that these free or symbolically paid autonomous driving services are more inclusive and commercially experimental. As He Xiaopeng said, the reason why Luobo Kuaipao can break even at present is that "many accounts have not been calculated yet."

After normalization and industrialization, plus various operating costs, it is still unknown whether normal charges can still be more cost-effective than human drivers.

Therefore, it is too early to worry that driverless cars will take away the jobs of human drivers.

Before full commercialization, will self-driving cars only cause controversy and waste money without bringing value to cities and us? Of course not.

One day in 2021, I interviewed the head of a domestic autonomous driving company and asked him: What are some of the technical goals you want to achieve within 1 to 3 years?

He said: We hope to achieve truly unmanned autonomous driving, without safety officers at all, and to achieve large-scale operations on open roads. This requires that every part of perception, decision-making, prediction, and control must be 99.99999 (seven nines) safe, which is the problem we want to solve.

Humans rarely or even basically do not participate in the driving of the car, which belongs to the L4 level of autonomous driving. In order to achieve this phased goal, the industry and academia need to make various preparations

Unmanned vehicles must have terminal-side intelligence and be able to be taken over by the cloud in real time. This requires a strong digital infrastructure, including high-performance domestic computing power, high-precision laser radar, large uplink low-latency network, cloud services that support large-scale distributed data storage and computing, etc. An autonomous driving industry can bring demand and market to related industries.

The development of autonomous driving technology cannot be accomplished independently by a single enterprise or scientific research institution. It relies on the innovation of integration of industry, academia, research and application, as well as open source data and papers, to reduce the R&D costs and risks of the entire industry. This has also promoted the scientific research progress of domestic high-quality data sets, NLP, computer vision, multimodal algorithms, embodied intelligence and other technologies, creating conditions for China to keep up with the world's cutting-edge technology trends such as large language models and multimodal models.

In fact, many autonomous driving manufacturers have become intelligent driving system solution service providers for OEMs, obtaining commercial returns from the B-end.

In the closed-loop exploration from research and development to implementation, autonomous driving has brought development opportunities to a large number of related industries and enterprises, and released many employment opportunities, such as data set annotators, unmanned vehicle road testers, AI software development engineers, ARM-based automotive chip designers, digital cockpit entertainment interaction designers, etc.

The industrialization of a general technology, like alternating current and broadband Internet, will create immeasurable economic, social and environmental benefits.

Fundamentally resolving the conflict between cars and roads, autonomous driving can bring efficiency and civilization to cities, and return road rights and life to pedestrians. When pedestrians and drivers no longer have conflicts over road rights, let's reimagine urban travel and lifestyles from now on.