Wayve CEO shares its key ingredients for scaling autonomous driving technology

The co-founder and CEO of Wayve, Alex Kendall, sees the promise to put on the market the technology of his autonomous vehicle startup. In other words, if Wayve sticks to its strategy to ensure that its automated driving software is cheap to execute, agnostic hardware and can be applied to advanced driving, robotaxis and even robotics driving systems.

The strategy, which Kendall established during NVIDIA GTC conferenceStart with a learning approach based on end -to -end data. This means that what the system “sees” through a variety of sensors (such as cameras) translates directly into the way it rolls (like deciding to brake or turn left). In addition, this means that the system does not need to rely on HD cards or rules based on rules, such as previous versions of AV technology.

The approach attracted investors. Wayve, which was launched in 2017 and has raised more than $ 1.3 billion In the past two years, plans to grant its autonomous software to automotive and fleet partners, such as Uber.

The company has not yet announced automobile partnerships, but a spokesperson told Techcrunch that Wayve was in “solid discussions” with several OEMs to integrate its software into a range of different types of vehicles.

Its cheap software plan is crucial to win these offers.

Kendall said that OEMs that put the advanced driver system of Wayve (ADAS) in new production vehicles do not need to invest in additional equipment because technology can work with existing sensors, which are generally made up of surround cameras and radar.

Wayve is also “agnostic silicon”, which means that it can execute its software on everything that its OEM partners already have in their vehicles, according to Kendall. However, the Startup’s current development fleet uses the NVIDIA Orin system.

“The entry of the ADAS is really critical because it allows you to create a sustainable company, to build a large -scale distribution and to ensure that exposure to data can form the system up to (level) 4,” said Kendall on stage on Wednesday.

(A level 4 driving system means that it can navigate alone in an environment – under certain conditions – without the need for a human to intervene.)

Wayve plans to market its system at an ADAS level first. Thus, the startup has designed the AI ​​driver to work without Lidar – the light and television detection radar which measures the distance using laser light to generate a very precise 3D card in the world, which most companies developing level 4 technology consider an essential sensor.

Wayve’s approach to autonomy is similar to that of Tesla, which is Also works on an end -to -end learning model to supply its system and permanently improve its autonomous software. As Tesla tries to do, Wayve hopes to take advantage of a widespread deployment of Adas to collect data that will help its system achieve complete autonomy. (Tesla’s “complete autonomous” software can perform automated driving tasks, but is not entirely autonomous. Although the company aims to launch a Robotaxi service this summer.)

One of the main differences between the approaches of Wayve and Tesla from a technological point of view is that Tesla is only based on the cameras, while Wayve is happy to incorporate the Lidar to achieve complete autonomy in the short term.

“In the longer term, there are certainly opportunities when you strengthen reliability and the ability to validate a scale level to further reduce this (sequence of sensors),” said Kendall. “It depends on the product experience you want. Do you want the car to roll more quickly in the fog? So maybe you want other sensors (like Lidar). But if you are ready to understand that AI understands the limits of cameras and being defensive and conservative accordingly? Our AI can learn it. “

Kendall has also teased Gaia-2, the latest generative world model of Wayve, suitable for autonomous driving which forms its driver on large amounts of real and synthetic data in a wide range of tasks. The model treats video, text and other actions together, which, according to Kendall, allows the Wayve IA pilot to be more adaptive and human type in his driving behavior.

“What is really exciting for me is the human driving behavior that you see emerging,” said Kendall. “Of course, there is no behavior coded by hand. We don’t tell the car how to behave. There is no HD infrastructure or cards, but instead, the emerging behavior focuses on data and allows driving behavior that deals with very complex and diverse scenarios, including scenarios that he has never seen before during training.

Wayve shares a philosophy similar to the autonomous truck startup Waabi, which also pursues a learning system from start to finish. The two companies have emphasized the data scaling AI models that can generalize in different driving environments, and both count on Generative AI simulators To test and train their technology.