Dong Che Hui
2026-11-18 22:30
In the past few years, Mercedes-Benz, a traditional luxury brand, has not been very vocal in the field of intelligence. When new car-making forces carry out high-frequency iterations around chips, large models and smart cockpits, this century-old car company often appears to be slow due to long research and development cycles and cumbersome verification.
Entering the second half of 2026, Mercedes-Benz has significantly accelerated its intelligent layout.
Relying on the MB.OS architecture that decouples software and hardware, Mercedes-Benz has recently launched two important updates. The first is the first new AI smart cockpit developed solely for the Chinese market, and the second is a pilot-assisted driving system equipped with the Momenta R7 world model that supports "parking space to parking space".
The implementation of these two systems has their own emphasis. The new cockpit will be first installed on the long-wheelbase E-Class, while the latest R7 assisted driving system will first be implemented on the pure electric GLC and pure electric CLA, and will later be covered in 9 core models such as the new long-wheelbase GLE and the new generation Maybach S-Class.
Mercedes-Benz officially summarizes this round of intelligent upgrades as "assisted driving and smart cockpit are aligned with the industry's leading standards." However, Dong Auto Club's experience found that whether it is polishing the details of the software interface or the commuting efficiency of assisted driving, this system still leaves a lot of room for improvement.
But if you compare Mercedes-Benz's past products vertically, this is undoubtedly a huge leap.
Easy-to-use new car engine dilutes the traditional "Mercedes-Benz flavor"
Judging from the actual experience of the Dong Auto Club, Mercedes-Benz’s “AI intelligent cockpit system” independently developed by the Chinese R&D team has almost no faults in terms of smoothness and functional completeness, and even has certain advantages in some places.
The most outstanding one is its response speed.
You can feel how fast this machine is without looking through the parameter list. Whether it is the system interface or the navigation map, the screen refresh rate has reached 60 frames per second. Mercedes-Benz officials also gave a more intuitive set of performance data:
Naturally, we will not test this data item by item during the test drive, but from a physical perspective, what Mercedes-Benz said this time is true. The lag and lag that were common in overseas brand cars in the past have completely disappeared on this system.
In order to adapt to the daily usage habits of Chinese users, Mercedes-Benz has reorganized the menu hierarchy, and the classification of functions is much more reasonable than the previous MB.OS. At the same time, the local application ecosystem of the new car has also been completed. Commonly used applications are basically there, and one-click interconnection with Apple CarPlay and Huawei HiCar is also supported.
In addition to the basic fluency and application ecology, the biggest change of this cockpit is that it introduces richer scene-based and intelligent agent functions.
First up is the brand new Scene Assistant. In this set of car machines, Mercedes-Benz has disassembled the software and hardware functions of the cockpit into independent "atomic capabilities". Users can freely combine 11 types of trigger conditions and 15 types of execution actions, customize some scene-based functions according to their own needs, and experience a task master similar to an ideal car.
Of course, considering that the target users of the E-Class may not be good at tinkering with these things, Mercedes-Benz has also prepared some preset plans for them.
Secondly, there is the AI capability of the new car. The so-called AI smart cockpit naturally has some "AI" aspects.
Mercedes-Benz has upgraded its AI travel assistant for in-car navigation this time, which supports automatic congestion avoidance, multi-destination route planning and fuzzy intention recognition. It is also the first to be connected to the Amap Street Sweeping List.
The voice virtual assistant adds voiceprint recognition and memory board functions, which can remember the voice characteristics and personal preferences of different drivers, such as temperature, seat position, etc., as well as your food taste.
After being connected with offline local life services, Mercedes-Benz's virtual assistant does feel a bit like a "personal assistant". You only need to issue a command, and it can schedule multiple cockpit functions at the same time, or recommend restaurants that suit your taste.
Mercedes-Benz said that probably in January next year, this system will be upgraded OTA to version 2.0 of the Doubao AI large model.
However, when vehicle-machine interaction is fully aligned with domestic mainstream solutions, the problem of homogeneity also emerges.
In order to retain the original brand characteristics, Mercedes-Benz's design team has actually made some attempts, such as making the car model in the middle more refined, adding a frosted texture shortcut bar to the interface, and supplementing it with silky flowing light and shadow effects. The look and feel of the entire interface is quite refined.
The most pleasing part is the air conditioning and seat adjustment interface. Mercedes-Benz specially made the crystal texture like that in MB.OS in the past, which can be said to be "Mercedes-Benz" at a glance.
However, the visual modifications are still difficult to conceal the "trace of new power" of this set of cars. The luxurious feel created by Mercedes-Benz in the past through physical buttons and the exclusive MB.OS interface has been diluted a lot in this universal interaction framework.
How to continue to strengthen brand recognition in UI and UX while ensuring ease of use is the next topic that Mercedes-Benz needs to explore.
Furthermore, placing a 3D car model in the middle of the screen is already the answer for the previous generation of domestically produced cars. Independent brands have already turned to more efficient map desktops or function cards.
Mercedes-Benz is now putting a lot of effort into polishing the car models to be sufficiently refined, but it always feels like it's a little too slow.
The R7 world model has created a unique "human-vehicle co-driving" experience
Intelligence extends beyond the cockpit. The other end of the spectrum is assisted driving.
In the first half of this year, Mercedes-Benz introduced the Momenta R6 flywheel large model based on reinforcement learning, opening up the urban and highway navigation that "can be driven nationwide". In the second half of the year, Mercedes-Benz moved Momenta's R7 world model into the new pure electric GLC.
According to official data, this model is based on 17 billion kilometers of real vehicle mileage training, and has refined more than 168 million pieces of golden data for iteration. On the underlying architecture, it first masters the laws of physics through pre-training with massive real data, and then uses the world model for closed-loop simulation and reinforcement learning to improve the system's ability to cope with complex long-tail scenarios.
When it comes to specific functions, this upgrade has brought many new changes.
The first is the parking link. It begins to support cross-floor memory parking, and the maximum memory distance has reached 3 kilometers. Secondly, for urban road commuting, the system has added roaming cruise assistance, which can realize autonomous car following, lane changing and traffic light recognition even if navigation is not turned on.
More importantly, with the support of this system, Mercedes-Benz has become the first luxury brand among Momenta's partners to support "parking space to parking space". Whether parked on the side of the road or in a parking space, assisted driving can be activated with one click by shifting into D gear.
Judging from the actual experience of Dongchehui, after replacing it with the R7 world model, the maturity of this system has indeed reached a higher level than before. The frequency of takeovers on urban roads has been greatly reduced, and manual intervention is not required in most road conditions.
But behind the low takeover rate, this system also exposed an obvious shortcoming, that is, low commuting efficiency.
In order to ensure driving safety, Mercedes-Benz has set up a double "safe space" in the bottom calibration. On the one hand, it focuses on preventing pedestrians and non-motorized vehicles, and performs defensive driving at blind spot intersections. On the other hand, it greatly reduces unnecessary braking and successfully optimizes the feeling of heavy braking by 2 times.
This strategy that attaches great importance to smoothness and peace of mind has both advantages and disadvantages on the actual road. The advantage is that the passengers in the car feel very stable, but the disadvantage is that the system is too cautious when starting up, passing through complex intersections, or needing to take a detour.
Another area worth discussing is the classification of driving modes.
Mercedes-Benz has set up three modes for this assisted driving: "calm", "standard" and "agile". In calm mode, its current slow pace is very consistent with its positioning; however, after switching to agile mode, the vehicle's lane changing and detour strategies are still not aggressive enough.
In actual driving, these three modes are more like "very slow", "very slow" and "slow". In this way, the meaning of switching modes is lost.
Indeed, Mercedes-Benz customers in the traditional sense may be a little older, more calm in their work, and not so receptive to radical assisted driving. However, the pure electric GLC and pure electric CLA, which are the first to be equipped with this system, are facing a relatively young user group who value commuting efficiency more.
It would obviously be more reasonable if Mercedes-Benz could truly expand the radicalness of assisted driving by switching between three modes so that all users can find a comfortable way to use it.
In addition, Dongchehui also found that Mercedes-Benz still has some room for optimization in the interaction of assisted driving.
In the current version, the interface layout of the central control screen is exactly the same when the vehicle is turning on and off assisted driving. However, in actual driving, the functions required by the driver in the two situations are not exactly the same. For example, in assisted driving mode, you may need to adjust whether the driving mode is aggressive or conservative, modify the speed limit ratio, or set whether to pass the ETC when passing a toll station.
In current Mercedes-Benz cars, you have to click into a deeper menu to adjust these options, which is not particularly convenient to operate. In contrast, some new domestic models do better in this regard. After turning on assisted driving, the corresponding functions will be directly placed on the main interface, making it convenient for users to adjust at any time.
Having said so many shortcomings before, I am not belittling it. Although many joint venture brands are now using Momenta's solution, Mercedes-Benz still makes a memorable difference.
Faced with homogeneous competition, BMW chose to differentiate in the interaction of braking, while Mercedes-Benz chose "co-driving with people and vehicles." Their assisted driving system is the first in the industry to support human-machine driving in the full speed range of 0-130km/h.
Mercedes-Benz has set the triggering torque of the steering wheel very low. The driver can easily turn the steering wheel to control the direction at any time during assisted driving. The system will not exit abruptly, but will continue to monitor the hand grip torque and achieve millisecond recovery after the driver lets go.
Judging from the experience of Dong Auto Club, this is the best and most boring experience among all brands.
That’s not to say that its assisted driving capabilities are not good, so it requires human-machine driving. Mercedes-Benz's choice and approach are ultimately a form of respect for the driver, which is also the core value of the entire system.
Li Hua
car editor
Anyone with wheels is interested and welcome to communicate. Email: [email protected]
A total of 1041 articles have been published
This article comes from the column
Next article2026-11-19 01:16
"GTA VI" launch experience: The best game of the past decade, and may be the best game of the next decade
Previous article2026-11-18 18:45
Why do this generation of college students choose to use Hongmeng Innovation?
Aifaner App
Love style, make the future within reach
Digital trendy brands focusing on tomorrow’s products
Advanced tools, prophets first, AIGC’s inspiration guide
A new era of car manufacturing, tomorrow’s travelers
Explore new lifestyles in the city and be your guide to tomorrow’s life
Copyright © Guangzhou Liwo Zhiyuan Investment Management Partnership (Limited Partnership) 2008 - 2026. Use of Aifaner website content for commercial purposes requires permission. Non-commercial use is licensed under CC BY-NC 4.0.
All content is made available under the CC BY-NC 4.0 for non-commercial use. Commercial use of this content is prohibited without explicit permission.




