AI Division

Who are we?

Located in Haidian District, Beijing, FATRI AI Division focuses on artificial intelligence algorithms, sensor technology, functional materials technology, data collection and signal processing technology. Relying on the machine nervous system platform developed by FATRI LAB, AI Division has developed the medical artificial intelligence system and the automatic driving system.

Our capabilities

FATRI Artificial intelligence team focuses on realizing the unmanned vehicle system through artificial intelligence algorithm and sensor technology, which based on sophisticated sensor fusion design scheme and massive data training and verification of the vehicle. The automatic pilot system is divided into: sensing module, simulation module, decision making module, control module

The data around the environment can be obtained by various sensors installed around the vehicle, such as laser radar、camera and millimeter-wave radar. The real-time information such as location, category, speed and orientation of the participants can be calculated by multi-sensor fusion technology.

As an important component of the unmanned system, simulation possesses a large number of actual road conditions and automatic driving scene data. Simulation module are based on the large-scale cloud computing capacity, and builds the virtual operation capacity of millions of kilometers per day. Through the open simulation service, it can access to a large number of automatic driving scenarios, and quickly complete a series of work such as testing, validation and model optimization, covering a comprehensive and safe and efficient.

Equipped with a comprehensive prediction and decision making system, the automatic driving vehicle can make corresponding trajectory prediction and intelligent planning according to real-time road conditions and road speed limit, while considering the account safety and comfort to improve driving efficiency.

The interaction system of autonomous vehicle control and chassis has the character of precision, universality and self-adapt ability. It can adapt to different road conditions, speeds, models and chassis interactive protocols. The module’s control accuracy will reach the level of 10cm.