Trending Articles
Trending Topics
Trending
Amazon and Nvidia have launched new autonomous vehicle initiatives this month, as the tech giants seek to bring artificial intelligence to driverless cars.
AWS has announced a deal with automotive hardware company Aumovio, which was spun out of the German car parts company Continental last year. The agreement will see the pair work together on self-driving vehicle software. It will also see AWS become Aumovio’s preferred cloud provider for its autonomous vehicle work.
Aumovio hopes that, by integrating AWS’ AI into its workflows, it can cut the time it takes vehicle makers to develop and test their self-driving cars.
According to Amazon, the addition of agentic and generative AI to Aumovio’s workflows will enable engineers to more quickly get a sense of how self-driving vehicles perform in rare driving situations.
Before release, vehicles are tested against millions of different driving scenarios to see whether they drive safely. Engineers working on Aumovio’s platform can now use AWS AI to query the data in natural language to tease out how the vehicles drove in a given scenario, rather than having to sort through the information themselves.
“Now [engineers] can simply ask questions in plain language – like ‘show me all cases where pedestrians entered the roadway at night in rain’ – and get instant results,” said Amazon. “The AI helps surface critical safety scenarios and rare edge cases for further testing.”
The first project to use the Aumovio workflows with AWS is an autonomous truck by self-driving vehicle company Aurora. Aurora is hoping to start production of the driverless truck next year.
Nvidia has also been working on AI projects for autonomous vehicles. At this month’s CES conference in Las Vegas, the company released its Alpamayo suite of products for driverless vehicles, including open AI models, as well as simulation tools and datasets.
CES saw the official launch of Alpamayo 1, a vision language action (VLA) model that allows the systems behind autonomous vehicles to approach driving scenarios with what Nvidia calls “humanlike thinking”.
VLA models study visual and language inputs with step-by-step AI reasoning to decide on potential actions. Alpamayo1 has a one billion parameter architecture, although Nvidia expects to expand that with future versions of the model.
Alongside Alpamayo 1, Nvidia also debuted AlpaSim, a simulation framework for AI development which offers “realistic sensor modelling, configurable traffic dynamics and scalable closed-loop testing environments”, according to the company.
It’s also unveiled an open dataset of over 1,700 hours of driving data across what Nvidia describes as “the widest range of geographies and conditions, covering rare and complex real-world edge cases essential for advancing reasoning architectures”.
