news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

As China's automobile electrification rate exceeds 40%, the industry's topic quickly switched to "intelligence." Algorithms, big models, transformers... these artificial intelligence terms are gradually being used skillfully by automotive people.

Starting from the second half of 2023, several leading automobile companies have launched a series of intelligent driving promotions. Among them, Huawei's map-free solution, Ideal's 100-city NOA, and Xiaopeng's end-to-end have brought many impactful trend predictions to the market.

Tier 1 suppliers may be more accurate in understanding industry trends. Tier 1 suppliers have a close relationship with OEMs, and in their view, whether a certain intelligent function can turn predictions into reality depends on its “price-performance ratio”.

Jiang Jingfang, the current president of Zhixing Technology, has a nickname in the industry, called "Tier 1 Iron Lady". She has 20 years of work experience in established international Tier 1 companies, and then joined the intelligent driving solution provider Hodo Technology. In April 2024, Jiang Jingfang joined Zhixing Technology as president, and she claimed to have returned to the Tier 1 industry.

At the Yanzhi Auto Annual Meeting on August 1, Jiang Jingfang put forward many of her views on the intelligent driving industry.

She said that China's smart driving is very competitive, and the industry has introduced many technologies at a rapid pace, and one could easily fall behind if one is not careful.

She said that because of the emphasis on user experience, making intelligent driving is like making embroidery needles, and the performance needs to be constantly polished.

Jiang Jingfang said that she saw three major trends in smart driving, namely intelligence, overseas expansion and cabin-driver integration.

Intelligent

"Some people say that China's automobile industry will be electrified in the first half and intelligent in the second half. In fact, I feel that intelligence has been the main theme from the past to the present and in the future," said Jiang Jingfang. She believes that as the adoption rate of L1 and L2 vehicles continues to increase, intelligent driving will become an important car purchase criterion for Chinese consumers.

It is said that China's smart driving is very competitive. What is it competing for? Jiang Jingfang listed many examples: In terms of driving, the industry quickly launched high-speed NOA and urban NOA, and is now even studying memory driving that does not rely on high-precision maps. In terms of parking, the early AVP that combined the factory and the car side could not actually form a commercial closed loop, so the industry will "compete" in the real valet parking that does not rely on self-built maps and the factory side. In terms of algorithms, CNN was used at first. "If you didn't talk about BEV+transformer last year, you must have been out. Now, anyone who doesn't talk about end-to-end in the forum seems to be out as well." Jiang Jingfang said.

The automotive industry is experiencing a major technological revolution, with numerous scenarios for autonomous driving iterating rapidly. It requires large models, data annotation, and training clusters.

"I think that as the functions of autonomous driving become more and more similar, user experience becomes particularly important. That is to say, you need to have good human-computer interaction and more humanized decision-making. This is also an area that we will pay more and more attention to in the future," said Jiang Jingfang.

The basic driving solution is already standard for models below 100,000 yuan. Under the price war, the BOM cost is almost transparent. "In terms of basic driving, I think it is safety and going overseas that count," Jiang Jingfang judged.

High-speed navigation and automatic parking have also become the entry ticket for 150,000 yuan-class cars. The fierce market competition determines whether the car can be sold. "In this field, you have to have both high performance and cost-effectiveness," said Jiang Jingfang.

Although urban NOA has become a popular feature recently, and several car manufacturers have shown their practical results, Jiang Jingfang has some reservations about this feature. "In the future, urban scenarios will definitely require big data, big models, and big computing power. I think it is more appropriate for OEMs to take the lead in this area, because the data mainly comes from OEMs," she said.

In addition to driving solutions, Jiang Jingfang believes that parking solutions can stimulate an explosion in user experience.

In the early days, intelligent driving solution providers focused more on parking performance itself. "If others' parking efficiency is less than 45 seconds, then our OTA upgrade must be less than 40 seconds." Today, intelligent driving solution providers are aware of the importance of "human-computer interaction" and non-contact parking can bring consumers a very good experience. Jiang Jingfang said that intelligent driving solutions should constantly benchmark against some good companies and some good practices in the market to continuously improve their performance.

In terms of human-computer interaction, in addition to achieving an overall closed loop and automatic switching of multiple views, we also need to add more humanized and classical reminders, augmented reality, and more rendering of surrounding obstacles. Next, we can also achieve more integration between the cabin and the driver, which will make the entire scene more intimate.

Jiang Jingfang admitted that the development of intelligent driving solutions is like making embroidery needles, which means that the performance needs to be constantly polished. "Just like the parking experience, there are many areas that can be carved, and this is also where we are constantly improving." She said.

Car going out to sea and cabin driving in one

Jiang Jingfang admitted that it is quite challenging to export cars overseas. However, she believes that this will be the next growth point for Chinese cars. In addition, she believes that the automotive chips will also undergo systematic architecture changes, with the main trend being cabin-driver integration.

According to the China Association of Automobile Manufacturers, China's automobile production will reach 30.161 million vehicles in 2023; according to the General Administration of Customs, China's automobile exports will reach 5.221 million vehicles in 2023. Jiang Jingfang said that in 2024, China's automobile exports will rise to 6 million or even 7 million vehicles.

When talking about the "difficulties" of going overseas, Jiang Jingfang concluded that they are mainly three aspects: regulations, data compliance and local adaptation.

First, we need to meet different foreign regulations and adapt to local scenarios, such as meeting the needs of left-hand users.

Secondly, data compliance is required, so car manufacturers going overseas must invest in a compliance cloud and desensitize overseas data by removing license plates and faces.

In addition, users in each country have different driving habits, so intelligent driving solutions must be developed and optimized accordingly, and corresponding data collection must be carried out, and some special overseas scenarios must be continuously added.

Take parking as an example. In Saudi Arabia and other places, ceramic nails are used to make parking space lines to prevent drying. In addition, some overseas users are used to parking with the front of the car, and different places have different parking space identification for the disabled. In view of these situations, intelligent driving solution providers going overseas need to collect relevant data to avoid having to fly overseas for training.

"Although going overseas is a major trend, we must prepare ourselves to ensure that some of our domestic intelligent driving functions can be quickly adapted abroad," said Jiang Jingfang.

In terms of car chips, Jiang Jingfang believes that the previous cockpit, driving, and parking ECU structures will change, and cockpit integration is an obvious trend. She said that the advantages of cockpit integration are lower system costs, better communication, and better user experience. At present, Zhixing Technology is already doing preliminary research on cockpit integration.