On April 19, 2022, THE 2022 HAOMO AI DAY was officially held in Beijing. With the theme of "Accelerating to the Future", Zhang Kai, Chairman of Miller Wisdom Bank, and Gu Weihao, CEO of Miller Wisdom Bank, shared the latest strategies, technologies and product progress of Milliper, and announced a number of achievements and blockbuster releases: 2022 Milliper ranked first in China's mass production of autonomous driving. In terms of passenger cars, the driving distance of assisted driving users exceeded 7 million kilometers, and China's first large-scale mass-produced urban auxiliary driving product "Millima City NOH" will be officially listed recently; in the field of terminal logistics automatic delivery vehicles, Millipede Little Devil Camel 2.0 was officially released, with a bicycle price of 128,800 yuan, becoming China's first 100,000 yuan terminal logistics automatic delivery vehicle; technically, China's first data intelligence system MANA ushered in a number of heavy upgrades, and its perception ability improved by leaps and bounds, through the original "double stream" Perception model and self-developed BEV Transfomer, "so that China has no unrecognizable traffic lights and lane lines" has become a reality; cognitive ability, cost and evolution speed have also doubled, the cost of model training has been reduced by 60%, the acceleration ratio has exceeded 96%, and the automation rate of labeling AI has reached 80%, which has greatly reduced the cost of labeling costs.
With the continuous improvement of the penetration rate of high-level intelligent driving of passenger cars, the NOH intelligent pilot auxiliary driving system has gradually entered the city, and intelligent driving in urban scenes is becoming a new place of contention. From 2021 onwards, two generations of passenger car assisted driving HPilot products have been released and are being delivered in an orderly manner as planned. At the HAOMO AI DAY, The "NoH of the City of Millima" equipped with HPilot3.0 was released. This is China's first large-scale mass production of urban assisted driving products, the first heavy perception of urban assisted driving solutions, but also the first most practical and efficient urban assisted driving products in China in 2022.
At the scene coverage level, the system can realize the main functions of automatic lane change overtaking, traffic light recognition and control, complex intersection traffic, unprotected left and right turning in the urban environment according to the driving route provided by the navigation, and can also cope with complex urban traffic scenarios such as vehicle close entry, vehicle obstruction, intersection, roundabout, tunnel, overpass and so on.
At the hardware level, the system is equipped with an ultra-high computing chip with AI capability of 360T, cache 144M, and CPU computing capacity of 200K+ DMIPS; it is also equipped with 2 lidars, 12 Cameras, and 5 millimeter-wave radars, which maximizes the safety redundancy of the entire set of auxiliary driver perception systems.
In terms of product strength, the passenger car intersection pass rate equipped with the system exceeds 70%, the success rate of lane change exceeds 90%, and the traffic flow processing capacity is as high as 4. It is understood that in the daily test, the intersection pass rate of the "no end of the city NOH" is ahead of Tesla's performance in China.
At present, the "NoH of the Last City" has been deeply polished in Beijing, Baoding and other cities, and the vehicle equipped with the system will also be officially mass-produced in the near future. In the future, there will be more than 100 cities where the "NoH of The Last City" function will be implemented, and the number of passenger cars will exceed 1 million units.
In the follow-up interview session, Yiche also proposed the city version at the end of the day, in the case of extreme weather or unexpected conditions, can it be successfully completed? And how difficult is it to achieve urban NOH at night compared to daytime? How did you overcome it? question. The following answer is given:
To answer your first question, first of all, the current driving system is actually based on assistance, which is assisted by automatic driving or assisted driving, and it is impossible to cope with it like a person, or people cannot cope with all situations, and there will be system limitations. There are indeed some situations that the system can't handle, such as extreme close-range plugging, such as some tiny obstacles that suddenly pop out, such as a cat or a dog. Our system will iterate quickly, such as MANA-based cognition, when starting, if there is a big bus on the right, and to the traffic light it does not start, or it suddenly starts to obscure your vision. At that time, I can stop starting, or I can stop in front of its stop line and tentatively see what is happening next to each other. Because at that time, we were based on human driving experience, or human driving behavior in MANA to carry out rapid iteration and response. This is the first question to answer you.
The second problem is that the scene at night is actually far more complicated than during the day, because at night when we turn on ourselves, there are high beams, some very dazzling traffic lights, including searchlights, the blur and contrast of the night route, which are some difficult points in identifying and coping with. Different regions have different levels of wear, including some differences in our pavement. And this is the basic reason why we are now vigorously promoting MANA, that is, we want to use the whole set of MANA capabilities to quickly solve the problems of the industry. Only by doing so can we quickly push the noH out of the city.
At the end of the event, Gu Weihao said: "Miller is an AI technology company committed to autonomous driving, and technology is the unchangeable belief of Millide. All the possibilities of the future come from the technical cultivation of the people every day. Technological improvement is the tireless pursuit of the last person, every moment is refining. ”