![](https://static.metal.com/common.metal.com/images/header-en/downward.png)
![](https://static.metal.com/common.metal.com/images/header-en/arrow_right_black.png)
![](https://static.metal.com/common.metal.com/images/header-en/arrow_right_black.png)
![](https://static.metal.com/common.metal.com/images/header-en/arrow_right_black.png)
![](https://static.metal.com/common.metal.com/images/header-en/arrow_right_black.png)
![](https://static.metal.com/common.metal.com/images/header-en/arrow_right_black.png)
![](https://static.metal.com/common.metal.com/images/header-en/downward.png)
Beijing (Gasgoo)- On August 2, Lenovo Vehicle Computing announced its automotive-grade AD1 domain controller for the L4 autonomous driving market has rolled off the production line at its Hefei industrial base. This milestone establishes Lenovo as one of the first Tier1 automotive suppliers globally to successfully integrate the NVIDIA DRIVE Thor platform.
Lenovo Vehicle Computing’s AD1 domain controller boasts impressive AI acceleration capabilities with INT8/FP8 computing power reaching 2,100 TOPS. It meets ISO 26262 ASIL D standards for functional safety and complies with ISO 21434 for cybersecurity, offering a variety of sensor interfaces. Lenovo is developing this controller based on the NVIDIA DRIVE OS, an operating system designed for safe, AI-driven automotive applications.
The AD1 is tailored for L4 autonomous driving commercial scenarios, ensuring high computing power, safety, reliability, cost-effectiveness, and low maintenance. It paves the way for mass production of autonomous commercial vehicles such as Robotaxis, RoboTrucks, RoboBuses, and RoboSweepers. Autonomous driving solution provider and service operator WeRide has chosen Lenovo Vehicle Computing's NVIDIA DRIVE-based solutions for their next-generation autonomous driving technologies.
The increasing application of end-to-end large model technology in autonomous driving algorithms drives the demand for high-computing platforms. Lenovo Vehicle Computing’s high-performance intelligent driving domain controllers, incorporating the NVIDIA DRIVE Thor, use FP4 and FP8 calculations for AI inference, improving Transformer model inference speed by five times while maintaining high precision equivalent to FP16. Lenovo's self-developed AI middleware, UltraBoost, is also adapted for the AD1, enhancing platform computational efficiency in model acceleration, operator enhancement, and task scheduling, meeting the high computing demands of autonomous driving.
For queries, please contact William Gu at williamgu@smm.cn
For more information on how to access our research reports, please email service.en@smm.cn