On July 4th, it was reported that Jia Peng, the person in charge of the Ideal Car World model, has taken a vacation and his workstation is currently empty, according to Workplace Animal.
According to insiders, the current status of the person in charge is that he is on sick leave, but "this may also mean that he will leave his position soon." Another informed source said that Jia Peng may not rule out leaving his job to start a business in the future.
Jia Peng not only led the research and development of the VLA model for Ideal Automobile, but also was responsible for the overall pre research of cutting-edge technologies such as world models. At present, Xia Zhongpu, the end-to-end model manager of Ideal Automobile, has resigned and is preparing to start a business, except for the production manager who is still on duty.
The so-called end-to-end model, also known as "End to End (E2E)" in deep learning, refers to an AI model that can output the final result as long as the raw data is input. Applied to the field of autonomous driving, it means that with just one model, the perception information collected by cameras and other sensors can be converted into operational commands such as how to turn the steering wheel and how much to press the accelerator, allowing the car to drive automatically.
Unlike traditional rule-based intelligent driving assistance systems, end-to-end autonomous driving solutions mean that the entire process from perception to regulation is processed through advanced algorithms and deep learning techniques. The application of end-to-end technology in autonomous driving has transformed the architecture of combining multiple models such as perception, prediction, and planning into a single model architecture that integrates perception and decision-making.
Simply put, in the past, the autonomous driving route was like multiple people driving a car, while end-to-end technology is single person driving, which is closer to real human driving.
Ideal end-to-end research and development cannot be underestimated. According to a report by 36Kr Auto, in early 2024, the senior management of Ideal Intelligent Driving went to the United States to see Tesla's end-to-end intelligent driving effect and decided to push forward with the end-to-end solution at full speed. The team has built an end-to-end+VLM (Visual Language Model) technical framework. In April 2024, Ideal underwent a downsizing and adjustment of the company hierarchy, and the intelligent driving team was reduced from a thousand to several hundred people.
Due to the unexpected end-to-end performance plan, in October 2024, Ideal has fully pushed the end-to-end intelligent driving solution to users of the Intelligent Driving Max version.
The implementation of end-to-end solutions is a crucial battle for Ideal Auto to establish its reputation. Ideal Auto revealed during its Q2 earnings conference call in 2024 that it has internally launched a longer-term end-to-end VLA model (Vision Language Action Model).
The goal of the VLM model is to deeply integrate end-to-end driving models with multimodal visual language models (VLM) to achieve decision-making abilities closer to human drivers and achieve "driver agent" level intelligent driving, which not only enables autonomous driving but also understands user instructions through natural language interaction.
