How else can NOA be rolled out?

Looking back on the development of the intelligent driving industry in 2023 , the development of NOA technology has become one of the focuses of the industry. Major car companies and suppliers are focusing on opening up cities, reducing costs, and making the most of NOA.

At the same time, policies and regulations for smart driving vehicles have also evolved in 2023, which provides a legal framework and support for the development of high-end smart driving. The accelerated implementation of urban NOA is regarded as the harbinger of the advent of the L3 level autonomous driving era, and also heralds further competition and development on the autonomous driving track.

The word "involution" can well summarize the competition among urban NOAs in 2023. This trend has not ended. How can NOA involute in 2024?

Urban NOA, involved in "Kaesong"

Zhang Kai, chairman of Haomo Zhixing, said that competition in the automobile market has shifted from new energy to intelligent competition. Currently in the passenger car market, the penetration rate of L2 and above intelligent driving has reached 42.4% and will reach 70% in 2025. The introduction of intelligent products into cars is undoubtedly the key for car companies to win future product competition.

According to data from the Gasgoo Intelligent Driving Configuration Database, the number of domestic NOA functions installed in 2023 will reach 945,000 vehicles, most of which are concentrated in high-speed NOA, and the number of urban NOA functions will be 238,000.

Image source: Gasgoo

Looking at the development of urban NOA in 2023, it is inseparable from keywords such as opening up cities and reducing costs. Last year, major car companies continued to announce high-profile plans to open cities. In the first half of the year, they were still making plans in cities such as Beijing, Shanghai, Guangzhou, and Shenzhen. In the second half of the year, many car companies generally gave more than 10 times the opening plan. Target.

Yihang Intelligent said that the boundaries of the concept of "Kaicheng" are actually relatively vague. For example, if autonomous driving can only be done on certain sections of the city, does it count as Kaicheng?

The routes chosen by various car companies are not consistent.

Ideal experienced a jump back and forth between urban NOA and commuter NOA last year, but at the end of the year it was clear that the "official version" of the full-scenario intelligent driving NOA will cover national highways and ring roads and 100 cities.

NIO has taken a different approach, choosing to open roads by road, route sharing, and converging lines into a network to start the delivery of enhanced navigation assistance NOP+ in urban areas. On December 12 last year, Ren Shaoqing, vice president of NIO Intelligent Driving R&D, stated on NIO’s official App that in the Banyan 2.3.0 version, the enhanced pilot assist NOP+ was renamed “all-area pilot assist NOP+” and provided users with coverage on highways and urban areas. and a full range of piloting experience in battery swapping scenarios.

Huawei NCA, which claims to be “available nationwide” , is one of the most radical leading forces. At the Wenjie M9 launch conference held on December 26 last year, Yu Chengdong said that he would successively launch Wenjie M5/M7 across the country. As well as OTA upgrades to this function on models such as the M9 that will be delivered soon .

Recently, some AI TO Wenjie M5/M7 owners have received the V3.3.8 smart driving version, which brings the long-delayed "pictureless NCA" function. According to the official announcement, the first batch of car owners to receive the "No Picture NCA" push is only 200 people. The second batch of push will officially start on February 1 this year and is expected to be available to all M5/M7 users who subscribe to the ADS  2.0 high-end feature package. Owners of the Smart Driving Edition are being pushed out one after another.

Obviously, this wave of urban NOA involution will continue until 2024. At the beginning of 2024, Xiaopeng has already started a competition mode with Huawei.

On January 2, Xpeng Motors officially announced that from now on, the urban intelligent driving capabilities of Xpeng’s 243 city. The system version added to Kaicheng this time is Xmart OS 4.5.0, which has been pushed on December 28, 2023. Xpeng Motors said that in addition to expressways and main roads in these 243 cities, the coverage area also includes secondary roads and national highways between cities.

In addition, on January 23, Zhiji Auto City NOA also officially launched its push, with Shanghai taking the lead. According to official disclosures, the "Advanced Version of Urban Commuting Mode" that can be used nationwide will also start a nationwide public beta in the first quarter of this year.

Gasgoo believes that currently many long-tail scenarios need to be solved to achieve nationwide urban navigation. Covering consumers' daily driving routes through transitional functions such as enhanced LCC and commuting mode can bring a better car experience.

It is worth mentioning that whether it is Wei Xiaoli or Huawei, more than 70% of the models currently equipped with NOA pilot function as standard are mainly concentrated in the price range of more than 300,000 yuan, of which models between 350,000 and 400,000 yuan are the main force.

However, we can already see the emergence of mid-to-low-end models such as Xpeng P5 and Baojun Cloud. As models equipped with NOA functions gradually drop to the price range of around 150,000, the number of NOA vehicles installed is expected to exceed 8.5 million in 2025.

Besides "Kaicheng", what else can we fight for?

Yihang Intelligence believes that there are two key factors for the large-scale implementation of urban NOA.

The first is a data closed-loop system based on mass-produced vehicles. The complexity of urban scenes determines that the urban NOA solution must be a data-driven system. Therefore, large databases from mass-produced models and automatic and efficient massive data processing capabilities will be the cornerstone of urban NOA implementation.

The second is the mature application of BEV solutions. Due to inherent problems such as the freshness and cost of high-precision maps, the BEV-based solution that emphasizes perception and light maps has become an industry consensus.

From a technical perspective, since urban NOA is far more difficult than high-speed NOA, more powerful computing power and more advanced models are needed to deal with these challenges. Traditional L2 assisted driving mainly relies on 2D vision + CNN, while urban NOA relies more on the perception fusion algorithm processing of BEV + Transformer.

Since last year, based on the development needs of urban NOA, new domestic forces and suppliers such as NIO, Ideal, Xpeng, Haomo Zhixing, and Huawei have also vigorously explored the potential of the BEV+Transformer architecture.

From a cost perspective, high-precision maps have disadvantages such as high cost, low coverage, and slow updates, making it difficult to meet NOA's large-scale and rapid expansion needs in urban scenarios. Car companies have long talked about having no maps, and "map dealers" have also begun to streamline maps. Tencent's HD Air lightweight high-precision data and NavInfo's HD Lite and other lightweight map solutions have emerged in response to the situation.

It is worth noting that the heavy-sensing route generally requires more sensors to ensure data redundancy at the sensing level , which brings good news to lidar companies, but it also faces an increase in technical costs, so which route to adopt has not yet been determined. foregone conclusion.

There is another big variable. Will Tesla FSD be introduced into the country? And will it have an impact on market route selection after entering China? It also remains to be considered.

Zhang Kai believes that the evolution of technology will bring about the increasing maturity of intelligent driving functions, and also bring about the continuous decline of costs. For smart driving suppliers, only when their products achieve the ultimate cost-effectiveness can they enter the threshold of car companies.

At last year's HAOMO AI DAY, Haomo Zhixing released its second-generation HPilot smart driving product, which focuses on the "cost-effective" advantage and can meet the mass production needs of high, medium and low-priced models, namely HP170, HP370 and HP570. Even the "Haomo HP170" all-in-one parking and parking intelligent driving solution priced at only 3,000 yuan can also realize high-speed mapless (impractical high-precision map) NOH, autonomous parking, remote control parking and other functions.

Yuanrong Qixing and DJI Automotive are also working on cost-effective and low-cost NOA solutions.

Zhang Kai said that as the penetration rate of urban NOA continues to increase, intelligent driving will truly approach thousands of households. In daily driving, more than 80% of the mileage is completed in the city. With the implementation of urban NOA functions, more consumers will truly realize the convenience and safety brought by smart driving.

Chen Yuxing, founder and chairman of Yihang Intelligent, believes that urban autonomous driving must realize anthropomorphic driving through key technologies such as artificial intelligence , so that end consumers can feel the comfort and safety of the system, thereby enhancing consumers' trust in the system and inspiring The demand for more frequent use of autonomous driving functions will ultimately enable large-scale mass production applications.

Of course, safety is still the ultimate essential requirement. Li Auto Public Relations Director Yu Wei believes that safety is more important. In addition to the usable range, safety must be greatly improved.

The front station for high-end smart driving

Intelligent driving continues to develop, and urban NOA is a special stage in the middle.

In mid-November last year, China's four departments jointly issued the "Notice on Carrying out the Pilot Work on the Access and Road Access of Intelligent Connected Vehicles ", which formally set out specific requirements for the access specifications of L3/L4 autonomous driving and improved the relevant rules. . This means that my country’s high-end autonomous driving has clear policy support and definition of responsibilities.