Home > News content

Xiaomi's 500 -day industry insider: There is no surprise of autonomous driving technology

via:新浪科技     time:2022/8/13 23:02:54     readed:93

At the official announcement of Xiaomi (01810.HK) officially entered the 500 -day node of the electric vehicle industry, the chairman Lei Jun disclosed the latest progress.

On August 11, Lei Jun said in a public speech in 2022 that Xiaomi Autonomous Driving adopted a full stack self -development algorithm to invest 3.3 billion yuan in research and development in the first phase of the autonomous driving field. The overall size of the team is expected to exceed 600 people. At present, Xiaomi's autonomous driving technology has entered the test stage. The first 140 test vehicles will be tested across the country. The goal is to enter the industry's first camp in 2024.

However, the audience found a lot of shortcomings from the real shots of the autonomous road test displayed by Xiaomi. Some people in the industry pointed out that Xiaomi autonomous driving technology displayed in the video has no surprises and highlights.

Industry insiders told the Red Star Capital Bureau that the video showed the status of the test car more completely, covering the three major auxiliary driving scenes of "cities, high -speed, parking", but the autonomous driving technology can only be considered as the industry's conventional level. "Considering Xiaomi for 500 days, this progress is fast."

Industry insiders also expressed their concerns about Xiaomi's self -study strategy for Xiaomi's autonomous driving technology, and they are not optimistic about its goal of entering the first camp in 2024.

Question: Show the technology well

Xiaomi showed a video of a self -driving DEMO (demonstration), and the sharp -eyed audience raised a lot of doubts:

The parking distance in front of the red light is too far away, the road is lining up and waiting for the vehicle to be too far from the front of the car, entering the ring island without a left turn lamp, the vehicle's start -up response is too slow, the vehicle is not good in the road in the road timely……

Some car commentators pointed out on Weibo that some scenes are more deliberate and the vehicles are stiff, such as the automatic detour of the accident vehicle, the zebra crossing ceremony for pedestrians. And such laser radar (overall perceived hardware) specifications cannot get on mass production cars, and there should be many "modifications". The section selection is also relatively simple, obviously in the early stages. He also said that it is not easy to make such a demo without enough the number of teams.

Welll Smart CEO Chang Yufei bluntly said in the circle of friends: "Xiaomi spent so much money to acquire, the DEMO car made without surprise and highlights", "Apollo 5.0 open source version is just this level." Zhou Guang, the founder of Roadstar, Said, "I feel back to the Demo video in 2019."

The video shows the launch scenes such as non -protected automatic U -turn, automatic island detour, automatic downward ramps, and functions such as predetermined parking spaces, autonomous passenger parking, and automatic charging of robotic arm. Function, L4 -level automatic parking function, and leading driving functions of high -speed and urban road scenes.

"Two or three years ago, these autonomous driving and high -end auxiliary driving techniques will make people shine, but now it is the base operation of smart cars." Some independent car analysts told the Red Star Capital Bureau that Xiaomi is currently displaying it. The ability, there is no breakthrough achievement in today's smart car field. "The industry knows that this display video is nothing. The test car can see that it is BYD Han and BMW 5 Series. It is equipped with Xiaomi autonomous driving system, mainly collecting data. Sensors and other hardware are not necessarily the same as mass production cars."

He said that as a whole, the video showed the status of the test car more completely, and at the same time covered the three major auxiliary driving scenes of "cities, high -speed, parking", but only reached the industry's conventional level. "However, considering that Xiaomi has only been building a car for 500 days, this progress is fast. In general, Xiaomi's current autonomous driving system needs to be polished."

Worry: not optimistic about entering the first camp in 2024

In his speech, Lei Jun said that Xiaomi's autonomous driving technology adopted a full -stack of self -developed technical layout strategies, and also shouted the slogan "The goal is to enter the first camp in 2024".

Full stack self -research means that car companies can master the core technologies in related fields, and have dominance in the research and development field, without need to use external forces; it also means that high costs require not only greater research teams, but also invest more funds to invest more funds. , Even to acquire companies to make up for their own shortcomings.

全栈自研自动驾驶的成本压力不小。8月2日,小鹏汽车(09868.HK/XPEV.US)与阿里合建的自动驾驶计算中心落地,CEO何小鹏表示,智能化自研路线下,“到2025年小鹏汽车每年的算力费用会超过10亿元。”

The Red Star Capital Bureau has noticed whether the full stack of self -research must be based on the conditions and strategic goals of the car company. /Nio.us) NAD, Xiaopeng's NGP, Tsla.us FSD.

Xiaomi Motors' autonomous driving team currently covers the full -stack technology related talents such as sensors, chips, perceptual control algorithms, simulation technology, high -precision maps, high standard positioning, tool chain, and training capabilities. Xiaomi also wholly acquired deep -moving technology in autonomous driving startups, spent more than 2 billion yuan to invest in more than ten upstream and downstream enterprises in the field of autonomous driving, involving autonomous driving, core sensors, core actuators, and domain controllers.

Based on this, Mi Fan expressed confidence in the self -developed autonomous driving technology of Xiaomi full stack to the Red Star Capital Bureau. In addition, Xiaomi started with software, and it has innate advantages in the field of data terminals. It is the accumulation in the intelligent field ...

However, the industry in the industry expressed respect and concerns about the choice of Xiaomi full stack, and it is not optimistic about the big cake of "entering the first camp in 2024".

Zhang Chao, an commercial vehicle autonomous driving algorithm engineer, told the Hongxing Capital Bureau that auto companies develop self -developed autonomous driving capabilities and need to have an open autonomous driving AI chip and mass -produced car. Based on this Driving data can finally "feed" bicycle intelligence.

According to the information disclosed by the annual report, Xiaomi Motors was listed in the first half of 2024. Zhang Chao questioned, "It is difficult to truly conduct autonomous driving capacity tests and data collection without mass production cars. Xiaomi Motors was listed in the first half of 2024, and Xiaomi targeted the first echelon in 2024. From the time of listing to the end of the year (less than a year in less than a year ) How many cars can you sell? There are a large number of modified test cars. Xiaomi first planned 140 test vehicles, how many test cars can there be? How many kilometers can I run in total? "

Independent car analysts believe that only vehicle and road collaboration can truly realize autonomous driving. Bicycle intelligence is the choice of the OEMs to master the core discourse rights of the industry. "It is difficult for the current AI level to accumulate by the big data accumulation of the vehicle itself. It is impossible to achieve completely autonomous driving in complex road conditions." He said that although it admires Xiaomi's full -study practice, Xiaomi is Xiaomi, but Xiaomi The car is only two years old. Without a lot of road test data support, even Xiaomi cannot guarantee safety.

Public information shows that as of now, Tesla's autonomous driving test mileage has 35 million miles (about 56.32 million kilometers), and Baidu Apollo has 32 million kilometers. "Wei Xiaoli" is also ready to enter the online car.

Red Star News reporter Wu Danruo

Editor Deng Yiguang Editor Yang Cheng

translate engine: Google

China IT News APP

Download China IT News APP

Please rate this news

The average score will be displayed after you score.

Post comment

Do not see clearly? Click for a new code.

User comments