The 4th Global Future Mobility Conference Held in Deqing International Conference Center on October 29-30, 2020

Not only well-known domestic and foreign automaker executives and automotive industry experts and scholars, but also representatives from relevant government departments attended the conference, which focuses on “mobility technology linking future cities”.

At the conference, Mr. Kai Wang, the Chief Technology Officer of Ideal Automotive, delivered a speech entitled “Intelligent Cars Under Digital Reconstruction: The Cornerstone of Future Mobility”.

The key points of his speech are as follows:

Hello everyone! I really appreciate the organizing committee for giving me this opportunity to share my views and outlook on future mobility.

About “Intelligence”

In terms of future mobility, changing people’s lifestyle habits is the most important concept, therefore, “intelligence” is crucial.

In the past few hundred years, humans have moved from the age of agriculture to the age of big industry, and are now gradually entering the age of intelligence. Why do I say that? Everyone can feel that, for example, many people have already experienced the convenience brought by mobile internet in their lives. Even in China, some lifestyle methods such as digital currency have completely changed our lives. For example, checks used to exist in the United States, but never appeared in China due to completely jumping over that era.

In my personal opinion, what is the most important thing for the era of intelligence? It is to map the physical world into the digital world, and through computer assistance, fundamentally improve overall productivity, which is a representative of increased productivity. In fact, we can imagine that some core technologies in this field, such as big data, data fusion, closed-loop systems, and flywheel iteration, are all key elements of the era of intelligence.

For us, for example, in the age of big industry, it was more about big machines such as steam engines and engines, production tools, and more focus on processes like waterfall development processes, assembly lines, and related keywords. However, after entering the age of intelligence, these things have undergone fundamental changes.

In fact, as we already feel, PCs and smartphones have gradually mapped us from the physical world to the digital world. Increased productivity is accelerated on this new digital platform. For us, intelligent cars are a new entry point for this intelligent revolution, and I believe it can bring significant changes to our lives.

Because a car is a mobile data center, it can use various sensors to perceive the world. Currently, intelligent cars are just a single point, but when you have new data architectures, as well as extremely complex sensors, and connect them all through the most important Internet and Ethernet, it can be envisioned that this point will become a network in the future, digitizing this physical world quickly. Therefore, a brand new world slowly emerges before people’s eyes, that is, the digital world.For the entire automotive industry, I believe that intelligent vehicles are particularly important participants at this core point. Intelligent vehicles are now facing the point that the entire industry needs to upgrade its services. As we can see, the entire industry is undergoing dramatic changes, and both OEMs, Tier1, and Tier2 are upgrading their services. For automakers, the most essential concept to truly achieve intelligence is to make users feel deeply connected to their cars. Previously, we used to say that we needed to test drive a new car before purchasing it. Now, the car should test you. Therefore, the industry is facing the question of how to become a master of martial arts who needs to be proficient in both internal and external skills.

I have listed several particularly obvious areas where users can experience the high intelligence of a vehicle firsthand. One aspect is from the perspective of automated driving, which is gradually transitioning from assisted driving to automated driving, and ultimately to unmanned driving.

The stage we are currently in is assisted driving heading towards automated driving. For example, both inside and outside the car are areas where intelligent innovation can lead to intelligent experiences and ultimately truly intelligent lifestyles, which can truly change people’s habits.

These two aspects are just surface modifications that everyone can feel, but there is also internal power which involves the renovation of the structure. As this event occurs in a new era, the entire structure must undergo a complete renovation. Later, I will go into greater detail about the reconstitution of product architecture, technology architecture, and data link architecture.

The last aspect is that for intelligent vehicles, most of it is the “end” side. However, in order to truly change people’s lives, we need the “end,” “manage,” and “cloud,” which is actually the three aspects of intelligent vehicles, intelligent transportation, and smart cities.

This is the ultimate goal, and there will definitely be interaction. We will talk about these four points separately.

Assisted Driving

It is an extension of the industrial age, established under stringent rules. For example, “ACC adaptive cruise control” or “lane keeping” that we often use now are rules that we develop in traffic environments, and the car can just follow them. However, in the case of automated driving, we need to use a lot of data to establish some multi-dimensional considerations. Sensors, computing centers, and other equipment must come into play to really achieve automated driving.

When it comes to unmanned driving, we reach a higher level. This means that specialized artificial intelligence needs to evolve into generalized intelligence. I deliberately put automated driving in the first place because it is basically the brightest pearl on the crown of the entire automotive intelligence. It is a very complex system engineering, just like a barrel, which is difficult to implement.

The Importance of Data

During the entire development process, I believe that at this point, the importance of data is embodied. Going back to the concept of the intelligent era just now, we need to better map the physical world to the digital world and better model it there. Through the computing power of computers, we can fully display it, producing a disruptive effect. During this process, productivity will increase exponentially, and we will have completely different ways of playing.

The Hardware ComponentThe biggest problem facing the industry now is that after enjoying the new life experience brought by smartphones, people have high expectations for intelligence in transportation. However, there is still a gap for both sensors and chips as the core of operation, despite the rapid changes.

Software

From the algorithm perspective, breakthroughs have not been fully formed yet. For example, current AI algorithms used to be based on very complex models with a large number of parameters. Now, it shifts to MobileNet, a “lightweight network,” which is very small and does not require high embeddedness. We need to find a compromise. As we all know, the earlier this technology is implemented, the more self-driving cars will be on the road to collect data, and this can truly cause a significant change from quantitative to qualitative.

In this sense, as we move forward, it will become more difficult, especially for general artificial intelligence. Although AI can beat Go games, general-purpose computers are equivalent to the perception of a 4-year-old child in the physical world.

Let me share my thoughts on the transition from specialized AI to general AI. I personally believe that self-driving cars will be closer and closer to us, but truly unattended and unsupervised automation may take time with 5G technology. The important point is that the cognitive process is to perceive it first, then memorize it, and then abstractly associate and apply it repeatedly. This is a typical process of the intelligent era, and it also requires data and feedback to iterate like a flywheel to achieve a very intelligent effect, just like a child’s learning process.

This involves many aspects, and the difficulty lies in the transition from one-dimensional rules to multi-dimensional rules. At this point, how to deal with explosion of perceptions and data becomes a challenge.

For example, if there is a plastic bag in front of a car, according to one-dimensional rules, the rule is to brake or avoid obstacles. But if there is multi-dimensional perception or cognition, and the plastic bag is a soft object that can be crushed, everything changes. A new dimension is created, and identification and perception of this object become difficult. When we achieve general intelligence, it will definitely need a multi-dimensional supporting structure.The biggest difference between machines and humans lies in their multidimensional abstraction and association under different rules. This is actually the biggest constraint for digital computers to accomplish, and I personally believe that it may require quantum computers to solve, just like classical physics and quantum physics. Our digital computers implement the 0-1 logic, while quantum computers enable their states to be 0, 1, or both 0 and 1 simultaneously, due to the involvement of various states and quantum entanglement. The latter concept is very similar to the cross-domain connection of multidimensional knowledge structures. Therefore, I believe that true autonomous driving may need breakthroughs on quantum computers to be achieved in a real sense.

For current autonomous driving, people can define the so-called ODD, which refers to its scenario environment, which actually restricts its implementation. In some sense, autonomous driving is more like a digitized and rule-based transportation system, even though it is multidimensional. It may be realized when we come up with a breakthrough perspective.

Earlier we talked about the exterior of the car, and now about the interior. In fact, when we first entered the smart era, it was the mobile phone that led the way. You can imagine how explosive the data is on phones now. Currently, the sensations inside the car is mostly about the experience of smart convenience. Initially, everyone’s needs for cars were more like an extension of the smart phone era. I hope everything I can use on my phone can also be shown in the car. This was what everyone in that stage most eagerly desired. In fact, we have transitioned from smart convenience to smart experience, which mainly refers to a completely customized smart environment inside the car. It is like autonomous driving but creating a smart world. That means from the physical world to the digital world, more sensors are needed to help in hearing, seeing, even smelling, and so on, to construct this new digital world.

Eventually, when we see the dream life realized, I believe it will be just like the name of our ideal car company “Car” and “Home.” We hope our car can be driven like a car, and become your home when parked, as an extension of the third space. For people’s life, the first space is the space where you need to sleep, the second space is the working space, and the third space is everything else. And, I believe that in the future, as people hope to increase efficiency and completely change their lives, the weight of the third space will become greater and may even squeeze the first and second space. Our goal is to achieve that.

A large amount of data is needed to drive it.

For OEM, the whole system needs to change, and I particularly want to emphasize the first point of OEM’s internal strength, which is architecture.

This architecture is the common architecture in the enterprise level and can be divided into three particularly important topics:

• Product architecture;
• Technology architecture;
• Data-link architecture.Traditionally, from the perspective of the industrial age, the technical architecture is more like the EEA (electronic and electrical architecture) discussed by today’s car companies. The industrial process is more based on what I need and is decomposed into highly process-oriented tasks, especially implemented in the so-called waterfall mode. Its advantage is that many details can be finely thought out, but it lacks flexibility.

Approaching this industry in the intelligent era in the right way

Starting from the organization structure of OEM

They need to analyze demand, which firstly requires input from data (to be discussed later). In addition, how internal and external departments interact and how to efficiently complete the closed loop within the enterprise were mentioned earlier. I just spoke about some keywords in the intelligent era, which are critical for the industry to solve what changes in organizational structure should occur in this new era.

Technical Architecture

In traditional industry, it is impossible to completely abandon it in this era, especially for things involving high safety, such as autonomous driving and other vehicle control systems. You cannot completely separate from it, otherwise there will be danger. However, in the new intelligent era, it has been given new life and different demands. In addition to doing this, it is necessary to be very clear about how to connect and interact with the product architecture, with the front end able to connect to the product architecture and the back end able to connect to the data architecture.

Data Architecture

This is the most important and highest-tech part of the internet. How to integrate the outside world, including cloud, edge, and pipe, and truly complete the closed loop with the entire data chain, directly and digitally. This is a completely new challenge. In fact, for the entire chain, technical architecture can use tools from the industrial age, but whether it is the product or data chain, internet thinking is required.

For the interaction between intelligent cars, transportation, and smart cities, we believe that to build a bridge for the digital world, cars are a cornerstone and a mobile data collection station. However, cities also need to be established as fixed points to collect more information, and gradually achieve interaction to ultimately achieve truly intelligent living and change productivity. These changes will be profound.

Privacy issues in the digital age

Because data is the driving force of the entire intelligent era, OEMs and regulatory agencies may need to consider how to solve this problem and how to solve the de-identification issue. As a responsible car company, we hope to be a part of this discussion. From the perspective of Ideal Auto, our recent goal is to become a data-driven car technology enterprise. Data-driven determines how fast we can change, while technology determines our height. China’s intelligent automobiles have great potential because we have a highly efficient government, a very good innovation environment, the world’s largest market, and the most loyal group of users willing to embrace technology. Thank you, everyone!That’s all for Wang Kai’s speech at this conference.

This article is a translation by ChatGPT of a Chinese report from 42HOW. If you have any questions about it, please email bd@42how.com.