RoboTaxi in China Tackling Road Conditions 50X More Complicated than U.S.

Autonomous Driving System in China needs to handle “not-so-rare” situation of bicycles coming in opposite direction.
RoboTaxi fleet at Guangzhou China has increased to 100 vehicles.

Self-driving taxis have become part of the lives of citizens in Huangpu District and Development Zone of Guangzhou City. Starting from November 28 last year, WeRide (Wenyuan Zhixing), a Chinese self-driving startup, has conducted robotaxi (unmanned taxi) trial operations in these two areas covering nearly 145 square kilometers, serving nearly 5,000 passengers in the first month, completing 8396 Orders.

The latest operation data released by WeRide (Wenyuan Zhixing) on June 17, 2020 showed that the number of locally operated vehicles has increased from 20 to 40 in the first month. Under the influence of the epidemic, the order completion rate still increased by 2.5 percentage points, and the vehicle response time also increased from March. 45.8 seconds was reduced to 19.5 seconds in April.

WERIDE March-April Operation Data Comparison

Wen Yuan, who has worked in the IT industry for 24 years and served as the senior vice president of Cisco Greater China, knows that COO Zhang Li believes that the driverless industry will make him more excited and allow him to “work for another 10 years or even 20 years.”

Zhang Li’s imagination is that, Robotaxi is a “transformers” that transports people during the day and night. The Robotaxi taxi software not only serves passengers, but also is the front-end entrance of smart transportation. WeRide (Wenyuan Zhixing)’s “WeRide ONE” general algorithm for autonomous driving can achieve all-round and all-weather coverage of Robotaxi, including tunnels, bridges, highways, morning and evening peaks, nights, rainy days, etc.

The total mileage of autonomous driving has exceeded 1.5 million kilometers since last year. The round A financing is strategically led by Renault Nissan Mitsubishi Alliance. Currently, it is actively promoting the round B financing. Zhang Li believes that the profitability of Robotaxi should be viewed rationally, “all Robotaxi business models are pseudo-propositions without removing the saftey driver.”

Weride COO Zhang Li

“Transformers” that carry people during the day and transport goods at night

Below is an interview with WeRide COO Zhang Li at their Guangzhou Headquarter.

Q: What is the current situation of WeRide (Wenyuan Zhixing)’s Robotaxi operation?

A: WeRide (Wenyuan Zhixing)’s driverless fleet has more than 100 vehicles, 40 of which are used as Robotaxi, and 60 vehicles are subject to special tests. As the scope of operations and order channels expands, it will be flexibly scheduled in the future.

The Robotaxi we operate is a compliant Guangzhou Baiyun taxi, and the billing is also based on the Guangzhou taxi standard, with a starting price of 12 yuan and 2.6 yuan per kilometer. In actual operation, dynamic adjustments will be made through the distribution of free monthly cards and other activities, but it will only be lowered.

Through cooperation with shareholder partners (traditional taxi companies), all our safety officers currently have qualification certificates for Guangzhou taxi drivers. Every vehicle in the transition stage needs a safety officer and personnel management, which is a great challenge for the technology company. The shareholder partner brings us a lot of experience in vehicle management, vehicle maintenance, and personnel management.
In the case of a security officer, Robotaxi may not necessarily earn more than a taxi. It’s like Didi has paid 70% of the cost to the driver. If he really wants to make money, he must remove the safety officer. Now we still do more technical development, drive test data collection and improve the efficiency of running-in operation.

Q: The order and response speed in April has been greatly improved compared with March. How to do it?

A: The number of orders depends on the promotion of the advertising market, and the reduction of user waiting time depends on the improvement of transport capacity and system optimization. We took anti-epidemic measures in March and April, and there was no incident of infection. The App has launched a points mall recently this quarter, which has also increased user stickiness.

Starting in March, Baidu can reserve the Robotaxi service. In April, AutoX also announced that it will operate in Shanghai, which is equivalent to publicizing Wen Yuan Zhixing, because we are the first to operate autonomous taxis in the country.

Q: Traditional online car-hailing rarely mentions the indicator of order completion rate. What will Wen Yuan know about the behavior refer to this data?

A: Order completion rate is a very important indicator in Robotaxi’s operation, which can reflect whether the demand and capacity match. Robotaxi’s operation, capacity and coverage are all set, and cruise, order and driving routes are all completed by the system, which is different from ordinary taxis. Therefore, we will focus on how to complete the order within the coverage and limited capacity.

The order completion rate is still a reference value, and we are still exploring the best mode. Robotaxi was free when it started operations on November 28 last year, and passengers may include the “Wool Party” in addition to what they want to experience. After charging, the wool is gone, and there will be non-rigid users who come to experience it. We hope to cover users in Huangpu District and Development Zone who really have just needs, and slowly accumulate and stabilize.

Q: Why did you choose 8 am to 10 pm as the operating time period?

A: The earliest design and operation time will cover the morning and evening peaks to prove that Robotaxi can also cope with the morning and evening peak tests. Outside of this time period, the demand for public travel is reduced, which is not very cost-effective for us.

If I increase the operation from 10 pm to 8 am in the future, I will not consider focusing on manning, but will see if I can deliver the goods during this time. We also hope to operate 24 hours, but it depends on what is really needed. I call it a hybrid model. Road test + operation is a hybrid. Manned and cargo are also a hybrid. It may become a feature of Robotaxi in the future.

For example, in the morning and evening peak taxi demand is large, the manned mode is used. You can go to lunch box at noon, and you can also go to 711 and other convenience stores to deliver goods at night.
However, this is currently only an idea. Domestic manned and loaded are two different licenses. How to break through and define is still to be discussed. But in terms of business model, I think it is feasible.

WERIDE Robotaxi currently running in Guangzhou China

Q: How does WeRide (Wenyuan Zhixing) achieve zero accidents in half a year of operation?

A: Zero accidents are actually zero liability accidents. We have also encountered the situation of being rear-ended by others. First, unmanned driving technology can improve the safety of driving and reduce the randomness of driving. Secondly, we also made some signs to let the cars on the road know that we are self-driving vehicles and actively maintain a safe distance, such as adding some reflective strips or recognizable lights on the car.

Target 80% orders in 20% area

Q: When will WeRide (Wenyuan Zhixing) be able to open highway and rural road scenes?

A: After more than three years of research and development testing, as of the end of March this year, there have been 1.5 million kilometers of test mileage in China, and we have continuously improved the full-stack automated driving solution. At present, a set of general algorithms has been formed to serve Robotaxi in China Operation of the city. We are planning to build one of China’s largest autonomous driving computing power center.

Operating in 144 square kilometers in Guangzhou is actually a very good verification of this algorithm. We are not particularly eager to rush to the highway to operate, we must run into small streets and alleys. At this stage, we hope that we can get 80% of the orders in the 20% operating area, which is enough. After eating through a place, we can just copy the scene.

Q: When do you want WeRide (Wenyuan Zhixing) to open operations in Guangzhou?

A: Expanding the scope of operations in Guangzhou is an important development goal for us. In the next 3 years, you may see our operating area continue to expand. At the same time, the synchronization of policies and regulations is also crucial. We maintain close communication with relevant government departments to try our best to let the government understand the current progress and business models of driverless technology, but when it is not the specifics of the enterprise, it will not be the final decision.

Q: Does WeRide (Wenyuan Zhixing)’s WeRide Go’s current progress meet your expectations?

A: In terms of the number of users and frequency of use, I think this app is doing well. At present, I haven’t seen other homes operating in China using this independent app. However, in terms of seniority, only half a year, there are still many shortcomings, we are also adjusting and optimizing App functions through feedback from passengers. After all, unlike Didi, we develop an App with hundreds or thousands of people.

WeRide Go is for passengers, and it must also communicate with WeRide (Wenyuan Zhixing)’s automated driving system platform to achieve management scheduling and path planning, so it is not the same as Didi Travel’s software. At present, we are beginning to gradually integrate the vehicle-road collaboration system, which means that there are many things beyond the line of sight, including traffic information, flow, and traffic lights that are not visible to the human eye. WeRide Go is a very important entry point in this system, and our team is still studying how to better open the system. At present, there is still a certain gap between the App and expectations. Our background will be connected with smart transportation in the future.

Q: Can the domestic operating experience be directly used abroad?

A: At first, domestic companies learned more advanced technology and experience from the United States, but after more than two years, we found that the differences in traffic conditions, culture, driving habits, policies and regulations between China and the United States are relatively large.

The complexity of Chinese road conditions is definitely a big challenge for autonomous driving, but it is also an advantage because it can help us quickly see more long-tail scenes and then iterate quickly. The drive test efficiency in Guangzhou is 30 times that of Silicon Valley, and the number of counter-direction bicycles encountered per mile is 60 times that of Silicon Valley.

WeRide Autonomous Driving Road Test Data data shows that the road conditions in Guangzhou, China is at least 50 times more complicated than Silicon Valley, U.S.

In addition, with the promulgation of this year’s new infrastructure policy and the implementation of the vehicle-road collaborative demonstration project, China has gradually evolved from single-vehicle intelligence to single-vehicle intelligence + vehicle-road collaboration to improve safety, which is also Chinese characteristics.

I think that domestic experience can be moved to a part of foreign countries, but it is not completely reproducible, because it is absolutely impossible for foreign countries to have the efficiency of China’s concentrated efforts to do major things to do new infrastructure and 5G.

Road test without safety officer as Key Milestone

Q: What is the current progress of WeRide (Wenyuan Zhixing)’s Series B financing? Is there a listing plan?

A: Series B is in progress, and will be announced as soon as possible after completion. I think listing is just a way to establish new financing channels in stages. Listing is not the final result, but it will definitely be an important stage. Regardless of whether we get money in the capital market through listing or other financing channels, we hope to be able to treat driverlessness as a business and give investors real gains.

Q: At what stage do you think autonomous driving is currently developing?

A: This year is actually a driverless travel service and the first year of Robotaxi. The development of the next 2 to 3 years will be the key point to determine whether the self-driving company can be truly normal, to see who can really go to the unmanned passenger operation with the safety staff removed. How to go from a passenger operation with a safety officer to an unmanned passenger operation, the intermediate process is the key point for everyone to really compete in technology and operational strength.

I divided Robotaxi into 5 stages. The first stage is to make an unmanned vehicle that can be operated in an experimental environment. The second stage is to obtain an autonomous driving road test permit, which can be tested on the road with a safety officer. The third stage is the operation of self-driving taxis to carry passengers. We started operations on November 28 last year. The fourth stage is to get a government permit, you can take out the safety officer to do driverless drive test, now only Waymo has this permission in the world, the fifth stage is the passenger operation of a truly driverless taxi. Some companies in the first three stages have been completed, and some are still in progress. The fourth stage is the most critical and truly tough one.

Q: Will the autonomous driving company be truly profitable by the fifth stage?

A: Yes, if you don’t take away the security officer, bluntly all these business models are false propositions. 70% of Didi’s income is paid to drivers. In the case of unchanged unit price, 70% of the income that was originally paid to drivers, I can now turn this part of the expenditure into gross profit through driverless technology. Which company can have a 70% gross profit? Therefore, the unmanned driving after removing the safety officer, this proposition really began to hold.

Q: Do you think Robotaxi will replace the traditional car-hailing and taxi?

A: It will be a mix and match model. This year’s L4 driverless technology development and business model are actually more regional businesses, so they will definitely be mixed with the way of driving. Even if L4 level autopilot is realized in the future, I believe that cars on the road will not become fully driverless cars overnight, just like the process of the development of the carriage to the car, there is a period of time when the carriage and the car are mixed. If it is completely replaced, I believe that at that time we will discuss when L5 will be realized.

Visit ChinaPotion Research Store.

https://chinapotion.myshopify.com

Deep Research on the emerging China Tech Companies and Trends