NIO’s epic update today is just a “hole” dug by Li Bin 3 years ago

NIO Day 2023 is still a NIO Day full of warmth, and it is also the most hard-core NIO Day. NIO released the world's first integrated hydraulic active suspension that combines intelligence and high integration, which is the "SkyRide Tianxing" Fully active suspension".

Li Bin on the stage spent a lot of time introducing the system, from software to hardware, from technology to functions, in every detail. But in the end, it was the ET9 that passed the speed bump carrying a water glass that conquered all the spectators at the scene.

Intuitive and shocking.

It’s a pity that we won’t be able to see the NIO ET9 on the road that integrates various “black technologies” before 2025. Naturally, we can’t know whether the performance of Tianxing Suspension is as Li Bin said, but NIO has recently been Launched an active suspension "Youth Edition" – 4D Comfort Navigation.

Last week, Weilai released the Banyan·Rong 2.4.0 version of the smart car system, which brings more than 50 new functions and optimizations, covering driving experience, cabin entertainment, active safety and other fields. Today's NOMI can even recognize "Move Xiao Li forward" is an instruction.

However, the new feature that attracts the most attention from users is undoubtedly the 4D comfortable navigation that can only be experienced in the 7 and 8 flagship series.

A "map" exclusive to NIO car owners

The surge of electronic technology in the 1970s and 1980s allowed electronic control of brakes and traditional systems. However, it was not until the large-scale application of sensing technology and computer control in the 21st century that cars truly entered the era of electronics and information.

The original active suspension system with a single function has gradually developed in a more refined and efficient direction, and it has become the CDC (Continuous Damping Control) that we often hear today. This system introduces more advanced sensors and control algorithms, which can provide more continuous and precise suspension adjustments based on road feedback, better balancing the vehicle, handling and comfort.

However, this also means that the CDC system's responses are based on current and past driving conditions, rather than predicting future road conditions.

In contrast, Mercedes-Benz's "Magic Carpet Suspension" (Magic Body Control) takes a more advanced approach. It uses cameras to pre-scan the road surface to identify bumps and potholes ahead and make suspension adjustments before the vehicle reaches those spots.

It is naturally better to be able to predict the future and respond in advance.

However, magic carpet suspension also has its limitations. In complex or low-light conditions, the performance of the camera is not satisfactory. Even if it is replaced by a lidar with better performance, it may still be unable to scan the road due to the obstruction of the vehicle in front.

NIO's approach circumvents the perception problem very well, and its secret lies in a map.

To be more specific, NIO's in-depth self-developed intelligent suspension control domain can not only sense the suspension information, wheel information and body information returned from the chassis in real time, but also use the body sensors to continuously record the bumps on the road. Form a 4D traffic layer with road surface information.

The 4D traffic layer is like a map with 6 types of road events, including uphill and downhill slopes, speed bumps, large and small undulations, and continuous bumps. Every 4 times the vehicle passes the same road section, a road event will be generated to record the type and characteristics of this road section. Before driving through this road section, the suspension will be prepared to adjust the height and actively adjust the damping control strategy in advance.

Xiao Baihong, head of NIO vehicle application software, said that as long as a NIO model equipped with air springs and CDC shock absorbers passes through the same road section 4 times, the bumpy layer selection of the road section can be confirmed, whether it is The same NIO car.

In this way, the freshness of NIO's "map" can reach "heavenly" level, and as more NIO vehicles support this function on the road, this "map" will become more and more complete.

In other words, all NIO cars with air suspension and CDC on the road are drawing this map together and enjoying the benefits it brings together .

A bit like Tesla before it.

In March last year, Tesla provided a similar, but more rudimentary feature to all Model System comfort. These data all come from Teslas on the road, including Model 3 and Model Y that are not suspended.

Everyone is for me, I am for everyone

What Weilai and Tesla do can be said to be a kind of swarm intelligence.

Swarm intelligence refers to a collective intelligence formed by a group of individuals through mutual collaboration and information sharing. Typical examples include social media, Wikipedia, open source software, etc. In the automotive industry, the most cutting-edge example is Tesla's FSD.

In 2022, Tesla held its first AI Day on October 1. When I knew the news, I was extremely happy. It meant that I could devote myself to work on National Day and work together with the motherland.

That day, Tesla's humanoid robot Optimus swayed onto the stage and gave the world a small cyber shock. At that time, Tesla said that the humanoid robot used the exact same computing platform as Tesla cars, and all its movements were learned through AI models.

The same goes for putting it in the car.

At that time, FSD Beta had more than 160,000 users. With the help of such a large "test team", Tesla already had more than 1.44 billion frames of video data at that time. At the same time, Musk abandoned the labeling team of nearly a thousand people last year, fully optimized the data labeling process, and increased the training speed by 30%.

On the same day, Tesla also introduced a term that we are now very familiar with, Occupancy. In this set of models, our world looks just like Minecraft, consisting of grids. Facing an "occupied" grid, even if the vehicle cannot recognize the obstacle, it will make avoidance actions.

At the same time, the vehicle can also recognize the slope of the road and make acceleration or deceleration actions in advance based on the actual road conditions. Tesla said at the time that FSD, which began to turn to AI, could list more than 75 million factors that affect decision-making in less than 10 milliseconds and quickly make the safest decisions.

Thanks to the addition of AI and a large amount of data feeding, FSD has grown rapidly.

Also last week, Tesla began to officially push FSD V12 (v12.1.2) to users. It wrote in the update description:

FSD Beta V12 upgrades the urban street driving stack into an end-to-end neural network. Millions of video trainings provided by the Tesla fleet replaced more than 300,000 lines of C++ code.

Overseas blogger Whole Mars (@WholeMarsBlog) said in his review video that the driving behavior of the FSD V12 is very natural. It will smoothly avoid vehicles stopped in the lane, change lanes decisively when necessary, and also face pedestrians. You won’t be as cautious as if you were frightened, and everything will go smoothly.

The blogger said that FSD Beta V12 "is the biggest advancement of FSD, achieving major improvements in all key aspects."

Nowadays, domestic independent brands are also working in this direction, and the "BEV+Transform" solution adopted by FSD has almost become standard.

At the Xpeng Motors 1024 Technology Day in 2022, Xpeng released XNet 1.0 based on BEV and Transform; the following year, Xpeng released XBrain, which consists of dynamic BEV + static BEV + XNet 2.0 that occupies the network. Huawei's ADS 2.0 is also based on the BEV + Transformer architecture and adds the GOD universal obstacle detection network, which targets the occupied network.

But in terms of data scale, Tesla is "far ahead."

Today, Tesla has sold more than 4 million cars; as of Q1 2023, FSD V11 has 400,000 subscribers; as of Q3 2023, FSD’s driving mileage has reached 525 million miles; Tesla’s Shadow mode can also gain driving experience without the car owner activating FSD.

Smart cars have expanded and developed from a car to a group. Weilai has obviously understood this truth very early, and they have been laying the groundwork for a long time to launch the 4D comfortable navigation function.

As early as 2021, when NIO ET7 was released, something called a 4D intelligent body control system also surfaced, which is the prototype of the current 4D comfort navigation function.

But this thing cannot be done by just one car. Weilai’s ability to fill the hole it dug today is inseparable from the efforts of all Weilai car owners here.

Anyone with wheels is interested and welcome to communicate. Email: [email protected]

# Welcome to follow the official WeChat public account of Aifaner: Aifaner (WeChat ID: ifanr). More exciting content will be provided to you as soon as possible.

Ai Faner | Original link · View comments · Sina Weibo