Wayve CEO shares his key ingredients for scaling autonomous driving tech 

Alex Kendall, co-founder and CEO of Wave, is watching his autonomous vehicle startup technology to market. This is, if the wave is cheaper to run its automatic driving software, hardware agnostic and advanced driver assistance system, robotaxis and even robotics are the strategy to make sure.

The strategy, which kept the kendal time Nvidia GTC ConferenceStarts with the last to the end of the data-powered learning method. This means that the system translates it directly (such as deciding to turn left or to turn left) through a variety of sensors (like camera). Furthermore, it means that the system does not need to rely on the HD map or rules-based software, as there are previous versions of AV Tech.

The approach has attracted investors. Wave, which was launched in 2017 and contained $ 1.3 billion raised over dollars Over the past two years, automotive and fleet partners have planned to give his self-driving software license UberThe

The company has not yet announced an automotive partnership, but a spokesperson told TechCrunch that Wave is “strong discussion” to integrate its software with multiple OMS.

Its cheap to software pitch is important to draw these deals.

Kendall said that OMS is keeping the Wave’s advanced driver assistance system (ADS) in the new manufacturing vehicles, no need to invest in additional hardware because the technology can work with existing sensors, which are usually made of camera and some radar.

Wave is also “silicone-magnostic”, which means that according to Kendall, the OM partners of the GPU can run the software on any GPU in their vehicles already. However, the current development of the startup uses the Nvidia Orin System-on-A-chip.

“Entering ADS is really criticized because it helps you get a sustainable business to create a scale distribution and training the system to get data exposure [Level] 4, ”Kendall said on the stage on Wednesday.

(A level 4 driving system means it in an environment – on certain terms – it can navigate it as it does without any human intervention)))))

Wave first plans to commercially commerce its system at an ADAS level. Thus, the startup is designed to work without the leader of the AI ​​driver – light detection and ranging radar that measure the distance to create a very accurate 3D map of the world using laser lights, which most companies that develop Level 4 Technology are considered as a necessary sensor.

Wave’s view toward autonomy is similar to Tesla, which It also works in the model of an end to end to strengthen its system and improve its self-driving software continuously. As Tesla is trying to do, Wave hopes to collect a broad rollout data from ADAS that will help its system reach the entire autonomy. (Tesler “Complete Self-Driving” software can perform some automatic driving tas

One of the main differences between the wave and the Tesla system from a technical point of view is that Tesla is only dependent on the camera, while on the other hand, the wave is happy to include the leader to reach full autonomy.

“Long -term, when you have the ability to create a level of reliability and to make it a level of shrinking, there is definitely the chance [sensor suite] Further down, “Kendall said.” It depends on the product you want to experience. Do you want to drive the car quickly through the fog? Then you probably want other sensors [like lidar]The But if you understand the limits of the camera on behalf of AI and are as a result protective and conservative? Our AI can learn it. “

Kendall Gaia -2 KO TOK, Wave’s latest generator world model is ready for autonomous driving, which provides a lot of training to its driver both real -world and synthetic data on broad tasks. The model processes the video, text and other actions together, which says Kendall allows the AI ​​driver of the Wave to be more adapted to its driving behavior and to be like humans.

“What is really exciting to me is that what is really exciting is the driving behavior that you see.” “Of course, there is no hand-coded behavior. We do not say how to treat the car. There is no infrastructure or HD map here, but instead, the emerging behavior enables data-driven and driving behavior that is very complex and deals with a situation that has never been seen during training and deals in various situations.”

Wave autonomous trucking startup has shared a similar philosophy with Wabi, which is also following an end to end learning system. Both agencies have given scaling data-powered emphasis AI models that can generalize Various driving environment across and depends on both Generator AI Simulator Give their technology tests and training.