The future of automotive technology will shift cars from simple modes of transport to ‘smart companions’, and innovators in the field, including Cerence, Geely, Li Auto, MediaTek, NIO, SoundHound, Waabi and Wayve, are increasingly harnessing NVIDIA’s cloud-to-edge technology to create intelligent AI assistants, enhance driver and passenger monitoring and boost general safety.
At the NVIDIA GTC convention, held in San Jose, NVIDIA is demonstrating its advances in generative AI, large language models (LLMs), and visual language models (VLMs) in the transport sphere. NVIDIA’s development of automotive assistants that boost safety, security and comfort, while merging deep learning and transformer models, is a key focus.
LLMs embody a significant aspect of generative AI and are comprised of transformer models; these are neural networks that excel at interpreting context and meaning. VLMs, another part of the generative AI family, possess language understanding and image processing capabilities, and unlike typical LLMs that are primarily text-focused, VLMs analyse and generate text via images or videos.
These technologies together empower NVIDIA’s Avatar Cloud Engine (ACE) and multimodal language models, working in conjunction with the NVIDIA DRIVE platform, to let manufacturers create their own smart AI-based in-car assistants. These can provide real-time assistance, personalised interactions and insights.
In addition, AI-aided surround visualisation can contribute to vehicular safety by sourcing broader information, such as local driving laws, and integrating it into decision-making. Cars will learn to adapt to drivers’ and passengers’ styles and needs, forging a highly personalised user experience.
Several of NVIDIA’s partners are also showcasing their pioneering work in generative AI using NVIDIA’s technology:
Cerence is collaborating with NVIDIA to enable deeper integration of CaLLM with the NVIDIA AI Foundation Models, a future project which will revolutionise in-car interaction.
Wayve’s GAIA-1, a generative world model for AV development running on NVIDIA is being exhibited, as well as LINGO-1, a closed-loop driving commentator enhancing the learning and explainability of AI driving models.
Li Auto’s multimodal cognitive model, Mind GPT, forms the backbone of their AI assistant, Lixiang Tongxue, allowing complex scenario understanding, knowledge retention and reasoning abilities.
NIO has launched its NOMI GPT, offering functional experiences such as NOMI Encyclopedia Q&A, Cabin Atmosphere Master and Vehicle Assistant for enriched user experiences.
Geely is targeting the creation of more intelligent, personalised and safer driving paradigms. They are using generative AI and LLM technology for intelligent navigation and voice assistants, among other things.
Waabi, utilising NVIDIA DRIVE Thor, aims to bring reliable autonomy to the trucking industry.
SoundHound AI uses NVIDIA’s technology to power its in-vehicle voice interface, consolidating real-time and generative AI capabilities.
Finally, MediaTek is set to reveal four automotive systems-on-a-chip within its Dimensity Auto Cockpit portfolio, offering AI-enhanced in-cabin experiences.