Source: Zhitong Finance
Author: Zhang Jiwei
On June 25, Xiaopeng Motor Chairman He Xiaopeng shared his self-driving experience with the Tesla FSD V12.3.6 version and Waymo in California. He pointed out, “Waymo performs better in downtown San Francisco, while FSD performs very well in Silicon Valley and Highways, can achieve very high scores, and handles many road conditions very smoothly.” He also said that he is very impressed that FSD has made great progress over the past few months, and Xiaopeng will also learn their excellent features and user experience parts from FSD. “I believe 2025 will be the ChatGPT moment for fully autonomous driving!”
Earlier, at the China Electric Vehicle 100 People's Conference Forum on March 16, Xiaopeng Motor Chairman He Xiaopeng predicted that the inflection point for advanced smart driving may come 18 months later. This will be a huge opportunity for Xiaopeng and other smart driving players to break the game in the knockout tournament. And how to face this knockout? He Xiaopeng said that the most important thing is to learn from the “big brother” in the industry and improve system ability, organizational ability, and management ability. However, he also suggested that “we cannot win this war according to the original approach.”
Meanwhile, at the company's 2024Q1 performance meeting, Xiaopeng Motor said that it originally thought that autonomous driving capabilities would be extremely strong in the second half of 2025, but now after the big models, this time may be brought forward to earlier next year.
According to reports, since AI Day on May 20, Xiaopeng Motor has fully promoted the AI Tianji system, which incorporates a large model, including a smart cockpit and intelligent driving capabilities, to the main models. If the use of high-precision maps is completely abandoned, the company expects that in the third quarter of this year, Xiaopeng Motor's XNGP will not only meet the national urban driving needs expected before, but also achieve the goal of being able to drive on every road. This will help Xiaopeng promote autonomous driving assistance technology more effectively around the world.
After the company's intelligent driving technology architecture is upgraded to be driven by an end-to-end large model, data barriers and data value will be significantly amplified. Through massive high-quality data training, XNGP will achieve or even surpass the performance of most average human drivers in complex traffic scenarios.
By 2025, the company plans to bring the urban smart driving experience to the highest level of high-speed driving at the current level, that is, to achieve a high level of autonomous driving that only needs to be taken over once for hundreds of kilometers in a city.
Editor/Jeffy