Alibaba and Nvidia Collaborate on Autonomous Driving
Alibaba Group's cloud computing unit and semiconductor giant Nvidia are collaborating on an artificial intelligence initiative that will allow Chinese automakers to develop a self-driving experience for smart vehicle owners.
The cloud computing services unit, the digital backbone of the e-commerce giant, announced the collaboration, unveiling a large-scale , multi-modal automotive solution developed in partnership with Nvidia and Alibaba's smart cockpit solutions provider Panama Network Technology, at the Apsara conference.
Alibaba Group's cloud computing unit has integrated its suite of large language models Qwen — including Qwen2-7B and Qwen2-VL — into Nvidia's Drive AGX Orin, a core computing platform for cars that combines advanced driver assistance, autonomous driving, and AI cockpit capabilities.
Large language models are the technology that powers generative AI services, such as ChatGPT.
This marks the first integration of large AI models from Alibaba Group's cloud computing unit into Nvidia's Drive vehicle platform, which is used by major Chinese electric vehicle makers, including Xiaomi, Li Auto, Great Wall and Zeekr, to power next-generation fleets.
NVIDIA's model acceleration technology reduces computing costs while minimizing latency in real-time processing of complex tasks with AI models from Alibaba Cloud, ensuring a seamless and uninterrupted smart experience for drivers and passengers.
According to Alibaba Cloud Services, thanks to Qwen's advanced capabilities in handling complex queries and visual intelligence processing, the solution based on the new large multimodal model can allow the in-car voice assistant to engage in dynamic, multi-role conversations.
The voice assistant can also make recommendations ranging from providing information about nearby landmarks to suggesting that the car's headlights be turned on proactively during certain conditions.
As part of a solution based on the large multimodal model, the Mobile Agent allows vehicle owners to effortlessly execute voice commands, such as ordering drinks through a food delivery app, resulting in rich in-vehicle experiences.
NVIDIA and Alibaba Cloud Services are pursuing a range of initiatives related to large language models for accelerated computing, which use parallel processing to speed up work and increase energy efficiency across demanding applications.