There is no war between the star startups in autonomous driving

The article is reprinted from Xin Zhijia, written by Shanshan.

After going through a period of frenzied and wild growth and enduring the capital winter, with the continuous improvement of technological maturity and the growing calls for mass production, China's autonomous driving industry seems to have entered the midfield of building momentum for large-scale implementation.

In recent years, the differentiation paths of many star autonomous driving startups that have taken the L4 Robotaxi track as their entry point are becoming increasingly clear.

Take Pony.ai , Qixing , AutoX , and WeRide, four Robotaxi startups that have attracted much attention in the industry, as examples. After several years of development and practice, they have each formed different understandings and views on the autonomous driving industry.

It is precisely these different views that have prompted these startups, which were originally on the same Robotaxi track, to take different paths.

Pony.ai, which has expanded the scale of Robotaxi operations, deepened its heavy-duty truck business, built its own standardized production lines, and may be planning to manufacture complete vehicles, may be the one with the heaviest operating model. Overall, Pony.ai has now chosen to focus its business on .

• Travel sector

Since July this year, Pony.ai has officially started regular operations in Jiading, Shanghai. Prior to this, the company has expanded its Robotaxi services open to the public in Nansha, Guangzhou and Yizhuang, Beijing.

To date, Pony.ai has expanded its footprint to the three first-tier cities of Beijing, Shanghai and Guangzhou in China, as well as multiple cities in California, U.S. Its fleet size has reached hundreds, and it has completed more than 295,000 orders worldwide.

According to Xin Zhijia, as the number of passengers experiencing Pony.ai's Robotaxi is increasing day by day, the company has set up a special team to be responsible for reception.

• Transportation

Pony.ai’s internal research and development for truck transportation (later named “Pony.ai”) began in 2018. Last year, Pony.ai first announced its research and development progress in this area, and soon after, it obtained the first autonomous driving truck test license issued by Guangzhou City, allowing it to conduct public road tests.

According to official introduction, Pony.com has conducted more than 200 days of road testing and has been recognized by two commercial operation partners. It has completed more than 13,000 tons of freight transportation and a commercial operation mileage of more than 37,000 kilometers.

Trunk logistics is a very large business, with a current output value of about 300 trillion yuan, of which road logistics accounts for more than 70%. At the same time, this industry has a large driver shortage and many safety challenges.

With the help of autonomous driving, the above problems can be alleviated to a certain extent; moreover, after achieving full unmanned operation, operating costs may be reduced by more than 40% - these are extremely attractive to logistics operators.

However, from the perspective of technology and large-scale implementation, L4 autonomous heavy trucks are no simpler than Robotaxi . Trucks have much higher requirements for safe braking distance than Robotaxi due to factors such as the variety of cargo they carry, large cargo capacity, and long body length, and therefore have higher requirements for long-distance perception.

In addition, if an accident occurs on a highway, the consequences of an unmanned heavy truck will be much more serious than that of a Robotaxi. If unmanned heavy trucks are to be driven on public roads, they may face greater resistance than passenger cars at the regulatory and policy level.

If trunk logistics, like Robotaxi, cannot achieve unmanned, large-scale commercial implementation of autonomous driving for a long time, then the dual-wheel drive business layout of "Robotaxi+Robotruck" is likely to bring pressure to Pony.ai's funding, personnel management, internal resource scheduling and other aspects.

At the same time, Pony.ai has previously built a standardized production line specifically for L4 autonomous driving software and hardware systems, which includes as many as 200 quality inspection items, including functional testing, environmental testing and stress testing, such as vibration, high and low temperature, waterproofing, noise, actual road testing, etc.

In addition, Pony.ai may move towards a more "heavy" model - not long ago, there was news that Pony.ai was about to join the ranks of car manufacturing, and its initial plan was to manufacture passenger cars. Although Pony.ai has denied the news, it would not be surprising if it eventually decided to start making cars.

Whether it is building a production line or manufacturing cars on their own, these actions are very rare among autonomous driving startups around the world. On the one hand, the operating costs are too high, and on the other hand, running such a large-scale business may be a big challenge for startups.

As the saying goes, if you want to wear the crown, you must bear its weight. We can wait and see how much potential Pony.ai can unleash in the future.

Compared with other players in the field, Yuanrong Qixing does not have too many self-operated Robotaxi vehicles. From this perspective, Yuanrong Qixing is undoubtedly the "lightest" company in the Robotaxi field so far.

After all, as a new mode of transportation, Robotaxi still faces multiple practical problems such as scale expansion, traffic acquisition, and commercial realization. Using all the power that can be used is undoubtedly the most convenient and direct way .

Based on this thinking, Yuanrong Qixing has previously mainly cooperated with OEMs and travel service platforms, with the partners providing vehicles and operating platforms and Yuanrong Qixing providing L4 autonomous driving technology to realize the implementation of the Robotaxi fleet.

The Robotaxi developed in cooperation with Dongfeng has landed in Wuhan and is open to citizens for test rides. The largest number of vehicles (30) were deployed in the first batch of the pilot project. The Robotaxi vehicles developed in cooperation with Cao Cao Travel aim to provide autonomous driving travel services for the Hangzhou Asian Games in 2022. At present, the two parties are conducting intensive road tests on the roads in Hangzhou.

This strategic model is also implemented in Yuanrong Qixing's cooperation in the field of commercial vehicles such as logistics. Taking the unmanned container truck project with Dongfeng Commercial Vehicle and Xiamen Yuanhai Terminal as an example, Liu Nianqiu once explained the role played by each partner in it: Yuanrong Qixing provides drivers and Dongfeng Commercial Vehicle provides vehicles.

From a business perspective, the asset-light model not only saves operating costs, but also enables a deeper understanding of the industry's specific needs for autonomous driving through cooperation with car companies and travel companies, thereby prompting players to fundamentally solve the problem of mass production of autonomous driving pre-installed systems and create products that are more in line with the market.

At this stage, for Robotaxi startups, the shared travel market is surrounded by wolves in front and tigers behind .

OEMs such as Geely not only have a large number of vehicles, but also have their own travel platforms; online car-hailing leaders such as Didi not only have mature operating platforms, but are also developing their own autonomous driving.

These companies have incomparable advantages over Robotaxi startups in terms of hard assets such as funds and vehicles, as well as soft assets such as operating experience and user traffic. Autonomous driving companies can achieve complementary advantages by cooperating with them.

It is worth noting that in addition to car companies and travel companies, Yuanrong Qixing has also recently cooperated with Futian District, Shenzhen.

On the 19th of this month, Yuanrong Qixing began to provide Robotaxi manned application demonstration services in the district, and has currently put 20 Robotaxis into service. With the government's approval and the in-depth development of operations, Yuanrong Qixing will gradually increase the number of operating vehicles, expand the operating area, and explore a more mature Robotaxi commercialization model.

It is reported that as of the evening of July 20, Yuanrong Qixing had received nearly 6,000 valid questionnaires, issued 100 invitation codes on July 19 and 20, and received about 100 trip orders every day. Starting from July 21, in order to meet the needs of more passengers, Yuanrong Qixing began to issue 300 invitation codes every day.

AutoX may be the only autonomous driving company in China that is currently focusing on developing fully unmanned Robotaxi.

Seeing that the large-scale commercialization of Robotaxi is nowhere in sight, peers have begun to try to explore other application scenarios. Only AutoX, which focuses on the "last few kilometers", is still struggling with the fully unmanned autonomous driving Robotaxi.

The reason why "full unmanned" is emphasized is that only by removing the safety officer can Robotaxi have the possibility of making profits . However, it is not easy to achieve this.

AutoX founder and CEO Xiao Jianxiong once lamented:

The L4 autonomous driving industry is just getting started, and many things in the product chain are not available for purchase... As a leading player in an emerging industry, the first to try something new even have to invent their own nets to catch crabs.

Previously, AutoX developed its own domain controller and vehicle wire control system based on this, and this is even more so with its investment in OEM Iconiq.

So far, AutoX has been operating the fully unmanned Robotaxi in Shenzhen for nearly half a year. If it wants to accelerate its scale in the future, fleet expansion is essential, and the node for mass production of its autonomous driving vehicles is approaching.

But for autonomous driving startups that don't have much funding, buying cars may not be a long-term solution.

Moreover, compared with retrofitted self-driving vehicles, pre-installed mass-produced models designed specifically for self-driving systems are obviously more reliable in terms of hardware stability and system adaptability . Especially for AutoX, which is committed to fully autonomous driving, its requirements for vehicles will undoubtedly be higher.

The fully unmanned Robotaxi fleet that AutoX has put into operation in Shenzhen is a rare type of self-driving vehicle in China, built on the basis of the Pacifica Grand Voyager. The Pacifica Grand Voyager was originally designed by FCA for Waymo specifically for self-driving.

Buying a car is not a long-term solution, but making a car yourself is even more difficult. Funding is one thing, and there are also many difficulties in terms of process manufacturing, supply chain, etc. At this time, investing in an overseas vehicle factory seems to be a good compromise option.

While investing in car companies, Xiao Jianxiong will also serve as ICONIQ's chief product officer, fully responsible for the definition and development of automotive software and hardware products and technical architecture. Xiao Jianxiong believes that this is an innovation in AutoX's concept, that is, AutoX can guide the design and production of mass-produced models based on its understanding of driverless technology, rather than having the car manufacturer define AI.

He said:

If a car company that does not understand AI produces L4 driverless cars, it will be difficult to manufacture cars suitable for the AI ​​era.

But can the fully unmanned Robotaxi route that AutoX has always insisted on really work?

Although AutoX has made many efforts in the field of technology research and development for fully unmanned autonomous driving, and recently launched the fifth-generation fully unmanned driving system AutoX Gen5 that is "born for unmanned driving", whether Robotaxi can achieve large-scale, fully unmanned operation on public roads ultimately depends on laws and regulations, and even ethical issues.

If the public cannot accept it and policies cannot be relaxed, then everything will become a mirage.

For AutoX, which is committed to creating fully unmanned autonomous driving, the road ahead may still be full of uncertainty.

On July 20, WeRide announced the full acquisition of autonomous driving trucking startup Muyue Technology. This acquisition may be seen as a signal that WeRide is officially planning its trucking business .

However, before this, as the first company in China to launch a paid Robotaxi service, WeRide, which has tasted success in the travel field, has always chosen to concentrate its firepower and continue to increase its travel services.

In 2019, founder Han Xu mentioned in an interview with Xin Zhijia that Robotaxi is a "technology + operation" issue. Based on this understanding, WeRide was already transforming from an autonomous driving company to a smart travel company at that time, and planned to develop Robotaxi as a product, focusing more on refined management and travel operations.

After launching the first fully public self-driving taxi service in Guangzhou’s Huangpu District and Development Zone that year, WeRide’s Robotaxi operating range has now been further expanded to core urban areas such as Guangzhou’s Haizhu District.

In February this year, WeRide also obtained a license to operate online ride-hailing services, becoming China's first autonomous driving company qualified to conduct online ride-hailing services. In addition to practice, WeRide's understanding of the future travel market may be the fundamental reason that drives it to move further and further in the direction of shared travel.

In the view of WeRide COO Zhang Li, sharing is an important direction for the future, and its value lies in " making C-end users no longer need to buy cars ", that is, changing the traditional mode of private vehicle purchase and optimizing the dispatch of traffic resources through shared travel. This will bring a subversive impact to vehicle retail, but it is an unprecedented opportunity for the shared travel market.

Based on this understanding of the future travel market, WeRide seized the public transportation attributes behind Robotaxi, and with the help of shared platform-based autonomous driving technology, it developed a business branch called Mini Robobus (L4 autonomous driving micro-circulation minibus).

This year, the pre-installed mass-produced L4 autonomous driving micro-circulation minibus developed in cooperation with Yutong has started regular testing in Guangzhou and Nanjing. In fact, both Robotaxi and Mini Robobus are unmanned manned transportation services, and the two landing scenarios have a high degree of overlap in business scope, technology, and safety requirements.

But in comparison, Mini Robobus, the so-called micro-circulation minibus, mainly solves the transportation connection problem of the "last few kilometers". Its routes and stops are relatively fixed, and at this stage it mainly operates in closed or semi-closed areas such as parks and scenic spots. The driving speed is relatively low, and the road conditions are relatively simple. The landing difficulty is simpler than Robotaxi.

Furthermore, in addition to the strong technical commonality between driverless minibuses and driverless taxis, the operational experience and team building of the two businesses should also be transferable and replicable. For WeRide, this should be able to reduce its burden in terms of personnel and organizational structure to a certain extent.

However, from the perspective of commercial implementation, Mini Robobus faces the same dilemma as Robotaxi .

As they are both unmanned manned services, even the slower Mini Robobuses cannot be driven on public roads before the regulations are relaxed, and can only operate in some closed or semi-closed parks. WeRide's Mini Robobus is currently only in trial operation on the International Bio-Island and Eco-Technology Island in Guangzhou and Nanjing.

Perhaps it is precisely because of this that WeRide now has plans to develop in the direction of freight.

Although these four Robotaxi startups have now taken different paths halfway through the race based on their early exploration and thinking, the tension is gradually fading.

But Robotaxi is its foundation. At the end of the game, all players will probably have to return to the main track of Robotaxi to compete with each other.

By then, can the experience points gained in different dungeons help them forge a winning weapon? It is worth looking forward to.