除前述外，一些主机厂的端到端智驾方案亦行至上车之前的准备期——小米端到端全场景智能驾驶于不久前开启定向内测、比亚迪成立前瞻部门主攻端到端技术落地、零跑B系列和C系列将于2025年实现自研端到端大模型智驾系统上车、奔驰自主研发的豪华品牌中首个“无图”端到端全场景智驾系统（L2++）计划随全新纯电长轴距CLA于2025年交付。The latest achievement of Ideal Automobile's intelligent driving end-to-end + VLM dual system architecture — the parking space function is officially fully pushed to AD Max users. As an independent brand focusing on smart automobile platforms under SenseTime Technology, SenseTime Jueying is one of the typical representatives of deep deployment of end-to-end intelligent driving technology in the supplier camp.
From the widely discussed 'Urban Area Without Maps NOA' earlier this year to the suddenly popular 'end-to-end' intelligent driving, will the field of autonomous driving witness a new technological revolution before the arrival of L3?
On November 28, Ideal Automobile's latest achievement in intelligent driving end-to-end + VLM dual system architecture — the parking space function is officially fully pushed to AD Max users. Users can voice activate one-click intelligent driving, gate control automatic start-stop, roundabouts, U-turns, construction zones, ETC, and other scenarios. According to incomplete statistics from Financial Link, the brands that have implemented end-to-end intelligent driving also include Tesla, Huawei's Quekun, and Xiaopeng Motors.
In addition to the above, some OEMs' end-to-end intelligent driving solutions are in the preparation period before implementation—Xiaomi's end-to-end full-scenario intelligent driving recently started targeted internal testing, BYD established a forward-looking department focusing on end-to-end technology implementation, Leapmotor's B-series and C-series are set to achieve self-developed end-to-end large-scale intelligent driving system implementation by 2025, and the Mercedes-Benz's independently developed luxury brand will introduce the first 'mapless' end-to-end full-scenario intelligent driving system (L2++) with the all-new pure electric long-wheelbase CLA scheduled for delivery in 2025.
As an independent brand focusing on intelligent automobile platforms under SenseTime Technology, SenseTime Jueying is one of the typical representatives deeply integrating end-to-end intelligent driving technology in the supplier camp. On November 27, SenseTime Jueying debuted the innovative large-scale intelligent cockpit product 'A New Member For U,' simultaneously releasing a range of advanced driving features covering high-speed, urban, and parking scenarios, demonstrating the mass production of Jueying's intelligent driving product system.
"General artificial intelligence models have brought three profound changes to various industries, including significantly increased production efficiency. The most typical example is end-to-end autonomous driving replacing traditional rule-based autonomous driving systems, where the work of thousands of algorithm engineers writing rules can be replaced by large model data-driven approaches." Wang Xiaogang, co-founder and chief scientist of SenseTime Technology and CEO of SenseTime Jueying, stated that in the field of end-to-end autonomous driving, SenseTime Technology's layout dates back to 2017. 'By 2022, after five years of research and development, we released UniAD, which has now been deployed in automobiles.'
According to the plan, SenseTime Jueying's full-scenario advanced intelligent driving solution based on the Horizon J6 platform will enter mass production in the second quarter of 2025, while the Jueying AD Ultra end-to-end mass production solution based on NVIDIA's Orin and Thor will be implemented in the fourth quarter of 2025.
The industry widely believes that with L3 level automated driving not fully available yet, the craze for end-to-end intelligent driving may trigger a new round of automated driving technology revolution. Among them, 'data' will become the most core asset and driving force.
The latest travel report released by Huawei's Qankun Zhijia shows that the total mileage of smart driving in October this year is 0.197 billion kilometers, and the accumulated total mileage of smart driving has exceeded 1.1 billion kilometers. The number of active users of smart driving in a single month is 0.3467 million, and the longest daily mileage per user reaches 2104 kilometers. As the market size of end-to-end smart driving models continues to expand, these real "mileages" will provide the most important support for the accumulation of effective data, continuous optimization and iteration, and long-term development layout by the enterprise.
"In Wang Xiaogang's view, "data accumulation" will become the key point to determine the initiative of autonomous driving in the future. When end-to-end autonomous driving comes, it will rely more on data drive. Whoever can be closer to the data will take the initiative in the future development process. "Various data collected during the autonomous driving process are very valuable when the vehicle behavior changes under various complex conditions," said Wang Xiaogang.
The latest report from the Car Hundred Think Tank under the China Electric Vehicle Hundred-Person Forum predicts that by 2030, the market size of the automotive data industry will approach 600 billion yuan. In addition, data security and compliance have become the key to building differentiated competitiveness for enterprises. It is estimated that by 2025, the market size of automotive data and cybersecurity will exceed 30 billion yuan.
