How to implement autonomous driving for commercial vehicles vertically

Sany Group Co., Ltd. was founded in 1989. Since its establishment, it has adhered to the corporate vision of "creating a first-class enterprise, cultivating first-class talents, and making first-class contributions" and created the well-known "Sany" brand in the industry. The mission of Sany Group is to "change the world with quality", that is, to change the world image of Chinese products with extremely high-quality products and services and create a world-class brand for the Chinese nation.

Dr. Wang Faping, Vice President of Sany Group, Dean of the Intelligent Driving Research Institute, and CEO of Haixing Intelligent Driving Technology, gave a speech on the theme of "Intelligent Networking - Reconstructing the New Future of Construction Machinery". The following is a summary of the speech:

Dr. Wang Faping, Vice President of Sany Group, Director of Intelligent Driving Research Institute, and CEO of Haixing Intelligent Driving Technology

Industry status and development trends

Sany Group has three main businesses. The first is the construction machinery sector, which has become the first in China and the second in the world. Some sub-sectors have become the first in the world. The second is commercial vehicles. Sany Group has currently ranked first in sales in China's new energy commercial vehicle sector for two consecutive years. The third is Sany Heavy Energy.

Next, we will introduce the current status and development trend of the autonomous driving industry. From the perspective of macro-trend development, first of all, the global urbanization evolution speed and China's urbanization speed are accelerating. China's urbanization rate was around 63.89% in 2020 and 64.72% in 2021. It is expected that around 2030, China's urbanization rate will reach 75%. The trend of urbanization is irreversible. Urbanization will bring changes in productivity, production tools, and production models. Construction machinery and commercial vehicles will inevitably be a sunrise industry.

Second, China's population is aging, and the number of working-age people is declining rapidly. The demand for talent and labor is a very big theme in China's rapid urbanization process, and the demand for robots/automated engineering machinery equipment will be even stronger in the future.

Therefore, the market potential of commercial vehicle autonomous driving is huge, with many players entering the market and hot industry investment, but where will the winners be? To have a winner, it is necessary to commercialize the many demands of commercial vehicle autonomous driving, such as small-scale transportation, fragmented park logistics, mining area transportation, and port transportation. How to realize these demands is the current dilemma of the industry.

Image source: Dr. Wang Faping from Sany Group

Commercial vehicle autonomous driving should be implemented "vertically"

In my opinion, first of all, fully driverless driving is the end game, laying eggs along the way can break the game, and large-scale productization can win the current situation. What should be done specifically? My first point is that lightweight reasoning computing, scenario-based software shelves, dynamic adaptive scheduling, and fast business closed loops are the only way to solve the industry's dilemma. First of all, lightweight reasoning means not to be obsessed with high computing power, but to focus on safety in the construction of the industrial chain, and to achieve efficient AI model application and generalization under limited computing power through methods such as AI model compression.

Second, build scenario-based software shelves. The scenarios applicable to commercial vehicles are different. Currently, there are six major mainstream application scenarios: port scenarios, park scenarios, mining scenarios, trunk logistics, airport scenarios, and municipal sanitation. In this case, it is impossible to have a set of software to cover all scenarios, and it must be customized based on the scenario.

Third, dynamic adaptive scheduling and rapid commercial closed loop. Construction machinery and commercial vehicles need to be scheduled during operation, and the scheduling level and speed in different regions are different. Therefore, enterprises need to have engineering development capabilities, build unified architecture design, unified coding specifications, unified API standards, unified deployment mode, and unified operation and maintenance services, which requires the collaborative efforts of the entire industry.

From the perspective of industry-wide collaboration, my second point of view is that application scenarios drive, industry partners collaborate, build and strengthen supply chains, and cultivate a large-scale industrial market. For China's commercial vehicle autonomous driving to be successful, it requires the participation and support of industry associations, and win-win cooperation can create an overall industrial chain.

Image source: Dr. Wang Faping from Sany Group

Sharing of Sany Group's engineering solutions and business models

Sany Group currently has several different scenarios in the field of commercial vehicle autonomous driving. The first is short-channel transportation in closed areas, the second is the overall solution for horizontal unmanned transportation in smart ports, and the third is the overall solution for low-speed unmanned logistics in smart parks. Sany Group's business in this area does not involve AGV automated guided vehicles and AMR autonomous mobile robots, but mainly focuses on logistics between factories, and can achieve connection in the fields of scheduling, communication, and postal services.

Image source: Dr. Wang Faping from Sany Group

For autonomous driving to be truly successful, the business model is crucial, and the core lies in making customers understand the content of the transaction. Sany Group’s business model is very simple, which can be summarized as “selling small boxes, selling medium boxes, and selling large boxes”.

"Selling small boxes" refers to the computing platform of Sany Group in the field of construction machinery and commercial vehicles. "Selling medium boxes" means that based on the current industry-leading concepts, the scene-oriented integrated perception software, decision-making planning and control software, central scheduling and OTA software are divided, and small algorithms that can be handled by dozens of lines of code are deposited in the algorithm library through API interaction and service call modes, so as to realize the centralized packaging of multiple algorithms, and then combine and apply them to specific scenarios for rapid integration. "Selling big boxes" means packaging functions such as vehicle-cloud collaboration, cluster scheduling and remote control, asset management, data-based predictive maintenance, network diagnosis and UBI data.

Finally, let me introduce Sany Group's thinking on the commercial vehicle autonomous driving industry, how to achieve large-scale commercialization of commercial vehicle autonomous driving, and how to provide turn-key services for N vertical industry scenarios through "one calculation, one software, and one cloud". Specifically, it uses a platform-based intelligent hardware base, a modular functional software shelf, and a cloud-native cluster scheduling platform to target different scenarios of commercial vehicle autonomous driving applications, and promote commercial realization through complete and replicable service solutions.