Making Logistics People More Dignified: How Can Autonomous Trucks Build a Next-Generation AI Logistics Network?

Guide

The national logistics hub and the national highway network constitute a strong logistics and transportation network, but there are still many pain points that need to be solved in this logistics and transportation network.

In the face of these pain points, Zhang Tianlei, CEO of Mainline Technology, said that traditional logistics has encountered development bottlenecks, and is currently facing many problems such as high operating costs of trucks, shortage of drivers, and frequent traffic accidents, and the "new generation of artificial intelligence logistics network" is the answer to solve these problems. The "new generation of artificial intelligence logistics network" will not only make the logistics and transportation network more efficient, but also make the front-line practitioners in the logistics industry more dignified.

Zhang Tianlei, CEO of Mainline Technology

(On December 17, Rogo held the "2021 4th China Logistics Science and Technology Innovation Enterprise Annual Conference and Logistics Capital Summit", this article is based on on-site shorthand)

First, about the main line of science and technology

Many people are puzzled by the name of our company and do not understand the meaning behind the word "main line".

In fact, this is a very programmer's characteristic name, in the work of programmers, the most critical program is called the main line, the rest are called branches, so we call the main line technology, this name carries our expectations.

Mainline Technology is a very typical self-driving truck company, and our team has been engaged in the research and development and application of driverless technology for more than ten years. In this process, we have made a lot of products and solutions that can be applied based on logistics scenarios, and we hope to connect our products with application scenarios and use digital means to form a new artificial intelligence logistics network.

In 2017, Mainline Technology was established, because of its excellent professional quality and the vast track it is in, the investors involved in our financing are industrial capital that can provide us with a lot of strategic resource support.

In the initial angel round, the artificial intelligence national team iFLYTEK gave us capital support and also gave us a strong industry endorsement. When the financing came to the A round, we wanted to solve the problem of mass production of autonomous trucks, and we were supported by NIO Capital and Bosch Venture Capital, the leading industrial capital in the automotive field. In order to integrate into the logistics industry, obtain resources, and accelerate the landing of autonomous driving, we found GLP, an industry giant in the field of logistics. After combining these resources with our own advantages, mainline technology has made great breakthroughs in technology, business, ecological cooperation, etc.

So far, we have achieved results that we are proud of in three areas:

1. Solid technology. We are now the only pilot unit of the Ministry of Science and Technology for the application of highway autonomous vehicles for trucks.

2. Leading scale. We have the largest commercial scale of autonomous truck operation clusters in China.

3. Ecological perfection. We have signed contracts with many logistics companies and fleets to actively build intelligent capacity networks.

In terms of team staffing, the programmers of mainline technology have formed a team that combines "old, middle and young", and Academician Li Deyi is our chief scientist. The core members of the team graduated from the Department of Computer Science of Tsinghua University. Tsinghua University has a lab with a more difficult-sounding name – the "Uncertainty Artificial Intelligence Lab" This lab focuses on the problems that can't be solved by automation in artificial intelligence, and for the past decade, we have been working in the lab to try to innovate in the field of intelligent perception, intelligent positioning and intelligent control in the field of autonomous driving, trying to make these innovative technologies available in real scenarios. Today, we are once again meeting in the main line of technology for this goal.

Second, a new generation of artificial intelligence logistics network

Our country is now in the process of upgrading from traditional logistics to digital logistics, and there are still many pain points in the existing logistics industry, so we put forward the concept of "a new generation of artificial intelligence logistics network".

China's transportation infrastructure is very good, there are hundreds of logistics hubs, connecting more than 160,000 kilometers of highways, these logistics hubs and the traditional infrastructure of highways constitute our rapid growth in logistics and transportation networks over the past 10 years. According to statistics, in 2020, the total amount of social logistics in the country will be 300 trillion, about 70% of which will be transported by trucks, and road transportation will still dominate.

The development of the traditional logistics industry mainly relies on the rough mode of stacking people and stacking trucks, and the production management is organized by experienced teachers, and we classify it as sweat logistics that relies on labor. Sweat Logistics has now encountered a bottleneck in development, with three main pain points:

1, high cost, fast growth. Now China's truck operating costs are very high, gross profit margins are basically in the single digits, and many individual drivers can only make a living.

2, recruitment is difficult, management is difficult. The number of truck drivers continues to decline, and the capacity is insufficient.

3. The risk is high, truck traffic accidents occur frequently, and the economic and safety consequences are serious.

Therefore, how to use technology to promote the intelligent upgrading of the logistics industry, which has many pain points, is the direction of our thinking. Can we use autonomous driving technology to digitally upgrade the transportation infrastructure that has been doing very well, and finally connect to form an intelligent logistics network.

To this end, we propose a set of frameworks for intelligent logistics networks, the underlying support integrates basic technologies such as deep learning, driving cognition and data closed loop, and then forms unique perception positioning and control technologies upwards, and finally forms two sets of solutions for closed scenes and semi-open scenes.

So can this intelligent logistics network finally be realized? In fact, China's high-speed rail network has set a good example, and we hope to create the same intelligent logistics network as the high-speed rail network. Theoretically, no matter which site you choose, I can get your goods on time and efficiently. This is a one-stop intelligent and efficient, on-demand billing, economic security and more dignified logistics network for practitioners, which we call "a new generation of artificial intelligence logistics network".

Third, self-driving trucks broke the situation

Facing the needs of real scenes, Mainline Technology has cooperated with mainstream commercial vehicle OEMs and suppliers to create three self-driving trucks, two of which are for logistics hub applications and the other for high-speed trunk logistics scenarios.

The basic principles of how self-driving trucks work are consistent with the way human drivers behave in driving. We have developed a L4 level autonomous driving hardware and software system "Trunk Master", which can be concretely understood as equipping trucks with invisible "artificial intelligence virtual drivers". In the process of controlling the vehicle, the human driver needs to perceive the surrounding environment, then form a precise positioning, and finally accurately control the vehicle.

In the same way, our "artificial intelligence virtual driver" needs to have perception technology, positioning technology, and precision control technology comparable to that of human old drivers in the process of controlling vehicles to complete tasks. Human beings have brains, and artificial intelligence virtual drivers have a "driving brain", which is a hardware product we make - L4 level automatic driving domain controllers, with powerful computing and task processing capabilities, which can replace human drivers in the car to complete the execution of actions.

As can be seen from the above figure, the world perceived by the self-driving truck is actually a high-precision world that reconstructs the real scene in three dimensions of the whole scene. We equip self-driving trucks with fully redundant sensor hardware, a complete set of camera-based cameras, lidar and millimeter-wave radar, ultrasonic radar, and multi-sensor fusion sensing solutions for combined navigation to ensure that our self-driving trucks can see farther, faster, and more completely, whether it is rain, fog, snow or night.

Today, after our extensive testing and commercial operation in real application scenarios, autonomous trucks have shown great commercial value and market potential in empowering logistics and transportation.

1. Intelligent and efficient. Self-driving trucks can not only release labor, but also achieve powerful intelligent management, thereby improving transportation efficiency and improving transportation quality.

2. Safe and stable. Self-driving trucks have multiple safety redundancy mechanisms, and technologies such as intelligent perception can effectively reduce the incidence of traffic accidents while achieving stable and long-lasting transportation operations.

3. Economic and energy saving. Self-driving trucks can not only "save people", but also optimize driving strategies through AI algorithms to achieve fuel savings. Judging from our current operating statistics, self-driving trucks have made very certain progress in this regard.

We hope to do a good job in the "technical work" that we are good at, so that our self-driving trucks rely on two scenarios (port logistics hubs and high-speed trunk logistics scenarios), as well as urban logistics scenarios that may be expanded in the future, to truly build a borderless autonomous driving freight network, that is, a new generation of artificial intelligence logistics network.

Fourth, empower smart logistics hubs

We have built two new energy autonomous trucks for the closed logistics hub scenario.

One is an unmanned electric collector card that retains the cockpit, which supports human-machine co-driving, which can not only completely unmanned operation, but also allow human drivers to drive this car to complete other tasks. Another ART artificial intelligence transportation robot without a cockpit is a mass production model customized according to the needs of port customers to build a new generation of intelligent container terminals, which can run all day long, drive in both directions, and can also drive obliquely and flexibly, more flexibly adapting to the container transportation operation process of the port terminal.

In order to facilitate port customers to better operate our self-driving trucks, we have also developed "One Cloud", that is, a cloud operation management platform, which allows customers to remotely manage vehicles in real time to achieve digital operation and maintenance. Previously, production operators had to manage vehicles on site, even in extremely cold winters, wearing thick coats, but now they can sit in a central control room similar to an airport tower with more dignity and comfort to complete the monitoring work.

Up to now, we have deployed more than 100 vehicles in the port logistics hub, which is currently the world's largest port unmanned truck cluster. In January 2020, we completed the delivery of 25 driverless electric collection cards in Tianjin Port, and in the same year, we delivered 13 unmanned electric collection cards to Ningbo Zhoushan Port. This year, we delivered 60 ART artificial intelligence transport robots and 6 driverless collection cards to the world's first zero-carbon smart terminal in Tianjin Port.

So far, we've done two lighthouse projects. Our driverless collection cards have helped customers transport nearly 500,000 boxes, and according to the customer's own statistics, the overall efficiency of the terminal operation has increased by about 30%. In addition, as a team of experts, we jointly applied with Tianjin Port Group and China Automobile Center to relevant ministries and commissions to support Tianjin Port as the only port autonomous driving demonstration area in China, setting a benchmark for the application of port automatic driving technology.

Fifth, the intelligent upgrading of road freight

In our trunk logistics scenario planning, we want to take the lead in opening up the Beijing-Shanghai line and the Beijing-Hong Kong-Macao line, because this is a very important golden trunk line in China, and then we hope to open up the Chengdu-Chongqing region, the middle reaches of the Yangtze River and the Yangtze River Delta urban agglomeration, and gradually form an autonomous freight network covering China's economically developed areas.

At present, Mainline Technology has built a fleet of nearly 50 trunk logistics autonomous trucks, and has cooperated with JD Logistics, Debon Express, Shentong Express and other partners to carry out special line transportation business, with a cumulative transportation mileage of more than 1.2 million kilometers and more than 1,500 unilateral cycles. Together with our partners, we obtained the first batch of open road test licenses for expressways in Beijing.

We've also created a Trunk Cloud operations service platform that supports fleet-level operations management. Through this cloud platform, operators can not only see the in-transit situation of autonomous trucks and monitor the operation status of the fleet in real time, but also realize a series of functions such as financial settlement and data statistics through the platform. More importantly, we work with the main engine factory to develop a data acquisition and analysis system in the forward direction according to a large number of terminal equipment deployed on the vehicle, through which we can know the status of every centimeter of the vehicle in real time, and can find problems that require early warning as soon as possible.

We are also currently focusing on the research of high-speed formation autonomous driving technology for trucks, which is the world's leading technology, which was originally only done by the European Union. We began to study this technology in 2019 and were selected into the 13th Five-Year Plan for Key Research and Development of the Ministry of Science and Technology. Now, our technology can support 4-5 trucks to automatically form a formation, the leading vehicle can be driven by people, can also be unmanned, and the accompanying truck can safely follow the leading vehicle at close range and driverless, which can greatly reduce costs and increase efficiency. This technique is also commonly known as "road train".

Sixth, the mission and road of the main line

In a sense, we are a very pure algorithm company, hoping that our algorithm can be made into a hardware product, and finally can be truly applied in the field of logistics. Today, there are about 10 million trucks a day, and 30 million truck drivers are running on the road, which is closely related to everyone's life. Therefore, we hope to use our own technology to bring some help to the entire logistics industry, but also to our society and people's lives.

China's high-speed rail network has become a well-deserved national business card, and we also hope that in the future, the "new generation of artificial intelligence logistics network" based on self-driving trucks will also be comparable to high-speed rail, making intelligent logistics a new business card of the country.

Author | Zhang Tianlei

Organize | Logistics salon

This article is the personal opinion of the author and does not represent the position of the logistics salon