Wayve approaches the Microsoft cloud to process data from autonomous vehicles

Wayve, a British start-up specializing in autonomous driving, announced on Wednesday May 18, 2022 a partnership with Microsoft to take advantage of its intensive computing infrastructure.

An essential point in the development of autonomous vehicles since these supercomputers can process large amounts of data. This announcement is not a surprise since Microsoft participated in the raising of 200 million dollars announced by Wayve last January and the American is multiplying partnerships in autonomous driving.

A win-win partnership

“Today a lot of supercomputer technology is about large-scale word processing or natural language processing”, Wayve co-founder Alex Kendall explained to VentureBeat. But for autonomous driving, it is essential to be able to process videos efficiently. Hence this partnership. Wayve insists this deal goes beyond the typical scenario where commercial app vendors partner with cloud providers. The start-up will work with Microsoft to push the limits of what is possible in its Azure cloud platform.

A win-win partnership is mentioned since the start-up will be able to help develop the infrastructure it needs and Microsoft will work on a use case allowing it to improve its platform in a sector in which the company is increasingly additionally present. Wayve assures that Microsoft’s technology will be able to process the more than one terabyte of data collected per minute by its autonomous vehicles.

Wayve’s different approach

Wayve explains that it has a different from its competitors allowing the autonomous vehicle to circulate in complex environments or on unknown roads only using cameras and a satellite navigation system. It is not necessary to embed high definition mapping or a large number of sensors. Its technology teaches how to drive like a human through imitation learning and reinforcement learning techniques. Learning by imitation makes it possible to copy human behavior and learning by reinforcement makes it possible to learn from the interventions carried out by security operators.

Its autonomous driving platform called AV2.0 constantly learns from driving data provided by partner vehicle fleets such as Ocado, Asda and DPD. Proving that its approach is working, the start-up has tested its London-trained technology in five other cities without first collecting data from those locations. Its autonomous vehicles therefore circulated in Cambridge, Coventry, Leeds, Liverpool and Manchester over a period of three weeks in September 2021.

Autonomous driving, a promising sector?

The autonomous vehicle sector seems to be attracting a growing number of players. Qualcomm announced in January 2020 that its automotive ambition goes beyond the digital cabin and in-vehicle connectivity. Since then, the announcement of partnerships has multiplied.

Similarly, cloud providers such as Microsoft Azure, AWS or Google are eyeing this market, which must process very large amounts of data and promote communication between vehicles and with their environment. Volkswagen and Cruise (the subsidiary of General Motors) have already announced that they are relying on Microsoft technology. And it is likely that the American does not stop there.

Leave a Comment