Li Xiang: Ideal car products and configuration core concepts to pay tribute to Apple

Home > Sci-Tech

Li Xiang: Ideal car products and configuration core concepts to pay tribute to Apple

2022-06-23 06:16:45 15 ℃

On the day before the morning market, the official APP was paralyzed and paralyzed, the ideal car held a media communication meeting on L9 smart driving, smart space, and vehicle safety. Essence

At the meeting, the ideal car CEO Li wanted to introduce the impact of Apple's product concept on the ideal product. Simply put, although the price of Apple products is slightly higher, the price is higher. On the one hand, Apple, through technology research and development, allows consumers to gain a higher level of experience, while equipped with a large number of software and technological innovations. On the other hand, through centralized selection and configuration, the delivery volume becomes very large and controls R & D costs. This is why the ideal L9 currently has only one model. The core concept of the ideal car for products and configuration is the true tribute to the choice of apple from the heart.

Regarding vehicle safety, Liu Liguo, vice president of electric vehicle electric vice president, introduced that the body of the ideal L9 has gone through 212 vehicle collision verification. 25%bias collisions in both sides are G (excellent level). And compared the BMW X7, the body to reverse the rigidity is 5%higher than the BMW X7, the lightweight is more than 8%, and the overall high -strength steel of the body accounts for 5%higher than the X7.

In addition, the technology of driving control in the body has been comprehensively improved, the suspension control system is fully developed, and comprehensive sensors, driving status, ADAS conditions, and comprehensive judgment of road conditions can be made. The suspension limit itinerary control, emergency variable line control, steady -state steering control, acceleration and deceleration control, etc.

Later, Lang Xianpeng, the vice president of the ideal car intelligent driver, brought information about the ideal intelligent driving. Lang Xianpeng explained why the supplier's purchase of intelligent driving was canceled. "The test process is long, the cost is high, the user feedback cannot grasp the ideal, the ideal wants to upgrade, the supplier cannot give it. Development. "

He also added that the ideal car builds a data closed -loop ability in terms of intelligent driving. For example, the test side develops an independent learning system, and no engineer does not need to find bugs in person. This is the core reason for the intelligent driving system that takes out of inferior friends in less than a year. In terms of research and development, the ideal also adopts the same "shadow" mode as Tesla. The development version of the intelligent driving algorithm and the current version decentralize user cars at the same time, which is more capable in the same scenario. Finally, Gou Xiaofei, vice president of the ideal car intelligent space, introduced the smart cockpit. Gou Xiaofei said: The way of smart cars should be natural interaction, that is, through the three -dimensional space interaction of voice gestures. It is unscientific to use the touch screen on the car, which greatly limits the "space" attribute of the car.

On the ideal L9, the "three -dimensional" is reflected in the more natural interaction. The voice assistant will appear on the screen and audio of the awakened area. It can also recognize the direction of people's fingers to replace the term "windows and sunny curtains".

5 interactive screens can drag interact with each other. And with different modes of synchronization, multiple perspectives in the same scene, and display separately. The core technical point is a multi -purpose visual fusion network based on deep learning. It can be identified with expression, gestures, and movements. The perceived accuracy is one millimeter.

In terms of voice recognition, the core is that MIMO_NET is used for deep convolutional neural networks used by multi -sound regions. The accuracy of the positioning of the sound zone increased by 20%. In the case of multiple people speaking, the identification error rate of the main sound area decreased by 30%.

In terms of interaction, the ideal builds a unique AI cognitive map. The same problem is targeted at different users, and the response given is different. "Ideal classmate" is not just a voice assistant in the future, but developing to robots.