36Kr Interview with tao ji, CEO of | Qianhang Technology: Autonomous driving technology and commercialization should be both hard-handed

Text | Angel Lee

Edit | Su Jianxun

The self-driving truck industry is ushering in a wave of new forces.

Recently, Tao Ji, CEO of Qianhang Technology, a newly established self-driving truck company in the industry, officially surfaced.

Founded in July 2021, Qianhang Technology is a company focused on L4 truck autonomous driving technology and commercialization. At the end of 2021, Qianhang Technology received 190 million yuan of financing from SF Holdings, Baidu, Xiaopeng Automobile and IDG Capital, becoming a new topic object in the industry.

After 12 years in Baidu, Tao Ji finally appeared as an entrepreneur. Tao Ji graduated from Xi'an Jiaotong University and Nanyang Technological University in Singapore, joined Baidu in 2010, and later became a member of the deep learning laboratory IDL.

When asked about the entrepreneurial feelings, Tao Ji laughed and said that this is the third time to start a business.

At the end of 2013, Tao Ji was responsible for starting the formation of an autonomous driving team, and after 2016, Baidu's Robotaxi (self-driving taxi) fleet made its first public appearance in Wuzhen, which was also led by Tao Ji. Since then, Baidu autonomous driving has entered an active development stage, and Tao Ji has also been active in the front line of Robotaxi technology, products and operations as the person in charge.

At the end of 2019, Baidu's Apollo platform was officially upgraded to three major business sectors: autonomous driving, automobile intelligence, and intelligent transportation. As an emerging business of Baidu IDG, intelligent transportation needs an experienced person to lead the team, and Tao Ji is the best candidate. After serving as the general manager of Baidu's autonomous driving division and the general manager of intelligent transportation product research and development, Taoji promoted the research and development and landing of a series of technologies and products such as Baidu driverless minibus/bus, vehicle-road collaboration, and intelligent transportation.

The experience of going from 0 to 1 twice also made Tao Ji, who shouldered the identity of the CEO of Qianhang Technology, feel relatively familiar.

The other two co-founders of Qianhang Technology have different backgrounds. Ding Fei was previously the head of the technical travel field of IDG Capital, and Sun Haowen was a member of the start-up team of the autonomous driving company Xiaoma Zhixing, and also the head of R&D of Xiaoma Zhika China, leading the truck team to complete the construction of technology 0-1.

After 5 or 6 years of development, the commercialization potential of self-driving trucks is being verified. Tucson Future, which has already been listed, has become the world's first autonomous driving stock. The joint venture company "Qingxiao Logistics Technology" established by Xiaoma Zhixing and Sinotrans was also officially established recently.

For the selection of the track, Tao Ji said that China's truck logistics industry has 7.3 million medium and heavy trucks, which is a trillion-level track, and the industry may have a severe labor shortage in the future, and there is a very large room for development in autonomous driving.

"In the next 10 years, there may be more than 700,000 heavy trucks that have not been driven, so it must be supplemented by means of science and technology and automation." Tao Ji said.

Image source: Qianhang Technology

Self-driving trucks replace human driving and provide logistics and transportation services, which is a big possibility.

First, the highway scene is simpler than the urban road scene. "The ODD (Design and Operation Area) of trunk logistics is a semi-closed scene, with no people mixed traffic and more complex intersection conflict points. At present, the total mileage of China's high-speed highway is 160,000 kilometers one-way, and many high-precision map companies have done high-precision map data for complete national high-speed. Tao Ji said.

In short, odd complexity is manageable. "Run these 160,000 kilometers well, and basically the problem of china's high-speed logistics automatic driving will be solved." Tao Ji said.

Second, the current truck logistics and transportation are basically equipped with two drivers, the labor cost is huge, and automatic driving is expected to gradually replace a driver to completely unmanned.

So, for the approaching end of automatic driving, how does Qianhang Technology consider it?

"We believe that we must no longer circle a small area to do demos, but aim at the goal of really landing products on a large scale." From the first day, Qian hang is to grasp both technology and commercial landing. ”

Qianhang Technology believes that the achievement of self-driving trucks requires technology, mass production and operation.

At the technical level, Tao Ji said that the team members have dual experience in the research and development of autonomous passenger cars and truck technologies. These experiences not only refer to the algorithm itself, but more importantly, the iterative thinking and data closed-loop experience summarized by the team in the past large-scale R&D and testing practices. "This kind of thinking and strong infrastructure support can be well migrated to the truck scene, which is a point that we feel very valuable."

But technology alone is not enough. Tao Ji said that after polishing the L4 technology, it must be deployed on a large scale, and "scale on the volume" is the most important. This also means that in the future, Qianhang will cooperate with the main engine factory to jointly create a truck fleet with reliability, consistency and stability.

In addition, with a fleet, how to provide logistics operation services is also very important. "To make a truck, you have to jump into the logistics industry. If the technique is in heaven, there must be two feet that can be inserted into the dirt. Only by going deep into the industry can we know the demand for technology in the scene, and can we truly open up the complete closed loop from technology to commercial operation. ”

Of course, as a new entrant to the industry, the top priority of Qianhang Technology is to build a team and underlying technology as soon as possible, and strive to catch up with the industry level. Tao Ji said that Qianhang Technology will open more than 100 jobs within the year, involving the pan-AI field, to autonomous driving, partial engineering infrastructure, hardware and mass production direction.

At present, the autonomous driving industry is still on the eve of the transition of technology to large-scale products, and although the industry pattern is not fixed, it is not without traces.

Although Qianhang Technology, which stands on the shoulders of its predecessors, has more opportunities for self-growth, it also means that it is necessary to cope with the challenges of products, business, capital and talents alone.

The following is an excerpt from the interview with Tao Ji of 36Kr (excerpted):

Media: From business executives to entrepreneurship, has the whole person changed now? What are the main aspects of the energy now?

Tao Ji: In fact, when I was in the big factory, the things I did were a bit like entrepreneurship in the big factory, a lot of innovation processes, the process of building a team from 0, this piece is relatively familiar.

Now the main focus is on two aspects: one is to build a team, more than half of the time is recruiting. The second is to take on the responsibilities of the CEO and ensure that everyone is aiming in one direction and moving forward quickly.

Media: Is it too late to start a business now? Where are the opportunities?

Tao Ji: Now the consensus is that the second half of commercialization has just begun, and it is not too late to join at this point in time. Our people are all veterans of the industry, and the benefit is that they know how to take fewer detours and how to quickly reach the target.

In the past few years, truck autonomous driving has only entered the fast lane of development. For example, when the truck was 18 or 19 years old, it was difficult for everyone to find a really easy-to-use original in-line control chassis, and even had to buy parts from a third party to modify it. In this way, it is difficult to expand the fleet on a large scale, which is the limitation of objective conditions.

And two years earlier, the self-driving function truck did not have a time point for mass production sops, and today we finally have a wire-controlled truck platform that may be good, but this car has just come out. The industry finally has a good platform to do technical iteration, and our time point is just right. At the same time, the SOP time point of the industrial chain and supply chain is about three years later, which is enough for us to polish out the leading automatic driving technology and complete the productization on the car.

Media: Can it be understood that this is a marathon, because the foundation is not perfect enough, so others have not run far, and Qian hang has a chance to catch up?

Taugi: I think so. Although today has not yet reached the end, but to the end of the thinking, that is, where is commercialization going? We mentioned this matter on Day one to be extremely important, which is conducive to our goal of really landing, not for the sake of technology.

Media: Qianyang is going to first engage in an L2 assisted driving car, first use it to collect data, and finally achieve L4 automatic driving?

Tao Ji: The route I proposed is a relatively complete technology stack, iterative method and infrastructure built by the research and development of large-scale autonomous driving of passenger cars, which is applied to the truck scene.

The amount of really valuable data that passenger car fleets can access every day is very large, and this data needs to have a good infrastructure platform to store, tag classification, and mining, turning it into fuel that can drive model and strategy iteration. We want to apply this method to the field of trucks, and use this data driven way to accelerate the development of the entire truck technology.

Media: What is the most important goal for 2022?

Tao Ji: Technology is definitely going to catch up in a big way, and the most important goal this year is to build a technical framework and talent team. The most important milestone is to validate whether the technical capabilities can reach a level that satisfies itself in combination with the first phase of business assumptions.

Media: Now that we are using fuel vehicles, will the future of using electric vehicles to do automatic driving of trucks be the direction you consider?

Tao Ji: Our choice of mass production platform is currently open, and we are also continuing to pay attention to the endurance of electric heavy trucks and the maturity of charging and replacing technology. One thing is for sure, we must be deeply cooperating with OEMs and Tier1, everyone has their own division of labor, and the automatic driving system must be able to reach the front of the car regulations and finally the volume.

Media: How to solve the problem of data backhaul with the main engine factory?

Tao Ji: On the one hand, OEMs need to work with partners to continuously improve autonomous driving capabilities, and OTA and data backhaul and sharing is a big trend. On the other hand, we will also build a fleet to participate in the operation, and the fleet data we operate ourselves will definitely form a closed loop.

Media: Are you going to build trucks?

Tao Ji: The company's goal is to do a good job in trunk logistics services. The truck is a production tool, and we use this tool to provide logistics and transportation services. Whether to build a car depends entirely on whether you can find a good tool and whether you can find a good partner to make a good tool together.

In the future, suppliers and OEMs may be partners, and we will define the vehicle together. In our cooperation with OEMs and Tier1, we will explore the boundaries of cooperation.

Media: The commercial model of self-driving trucks has not really fully worked out, what do you think of this problem?

Tao Ji: Although autonomous driving has not yet reached the end, it must have a clear commercial thinking and path. We believe that trunk logistics heavy trucks are the largest imaginary space, the clearest technical and commercial paths, and it is worth all in.

What we are doing is a product with L4 capabilities, but it shows different capabilities and application forms at different stages, scenario requirements and policy license conditions. We are very clear to see where the next technology mile stone is and what the corresponding business model is.

Media: Where's the next mile stone?

Tao Ji: First, from the product technology, we will verify whether the medium- and long-distance routes that originally required two drivers can be completed with the help of a single driver with the help of automatic driving.

But we must also admit that in such a track where technology changes rapidly and the business environment changes rapidly, it is still necessary to maintain the openness of the strategic direction, especially in the first one or two years to be able to respond and adjust flexibly at any time.

Media: Autonomous driving may have entered the second half of commercialization, from the technical thinking of five or six years ago to the current product thinking, what is the biggest difference between the two?

Tao Ji: A few years ago, everyone may have paid more attention to where the boundaries of technology are, where the technology can go, and that stage determines that everyone is focusing on the technology itself. In the second half of commercialization, everyone is entering the commercialization into the project. One of the biggest differences is that we pay more attention to the scene, pay attention to the real needs of the scene, and do not use technology to find the scene, but use the scene to pull technology.