Wayve’s co-founder and CEO, Alex Kendall, is optimistic about bringing his autonomous vehicle startup’s technology to the market, provided the company adheres to its strategy of developing affordable, hardware-agnostic, and versatile automated driving software. This software can be applied to advanced driver-assistance systems, robotaxis, and even robotics.

During Nvidia’s GTC conference, Kendall outlined the company’s approach, which involves an end-to-end data-driven learning method. This means that the system’s “vision” through various sensors, such as cameras, directly influences its driving decisions, like braking or turning. Notably, this approach eliminates the need for high-definition maps or rules-based software, which were essential components of earlier autonomous vehicle technologies.

This strategy has attracted significant investment, with Wayve, founded in 2017, securing over $1.3 billion in funding over the past two years. The company plans to license its self-driving software to automotive and fleet partners, including Uber. Although no automotive partnerships have been announced, a spokesperson confirmed that Wayve is engaged in “strong discussions” with multiple original equipment manufacturers (OEMs) to integrate its software into various vehicle types.

The affordability of Wayve’s software is a crucial factor in securing these partnerships. Kendall emphasized that OEMs can integrate Wayve’s advanced driver-assistance system (ADAS) into new production vehicles without investing in additional hardware, as the technology is compatible with existing sensors, including surround cameras and radar.

Wayve’s software is also “silicon-agnostic,” allowing it to run on any GPU its OEM partners have in their vehicles. Currently, the startup’s development fleet utilizes Nvidia’s Orin system-on-a-chip.

“Entering the ADAS market is critical, as it enables us to build a sustainable business, achieve large-scale distribution, and gather the necessary data to train our system to reach Level 4 autonomy,” Kendall stated during the conference. A Level 4 driving system can navigate its environment without human intervention under specific conditions.

Wayve plans to commercialize its system at the ADAS level initially. To achieve this, the startup has designed its AI driver to function without lidar, a sensor that most companies developing Level 4 technology consider essential. Instead, Wayve’s approach is similar to Tesla’s, which is also working on an end-to-end deep learning model to power its system and improve its self-driving software.

One key difference between Wayve’s and Tesla’s approaches is that Tesla relies solely on cameras, whereas Wayve is open to incorporating lidar to achieve near-term full autonomy. Kendall noted that the choice of sensors depends on the desired product experience, such as the need for the car to drive faster through fog.

Kendall also introduced GAIA-2, Wayve’s latest generative world model tailored to autonomous driving. This model trains the driver on vast amounts of real-world and synthetic data across various tasks, processing video, text, and other actions together. This enables Wayve’s AI driver to exhibit more adaptive and human-like driving behavior.

“What’s exciting is the human-like driving behavior that emerges,” Kendall said. “There’s no hand-coded behavior, and we don’t tell the car how to behave. The emergent behavior is data-driven, enabling the system to handle complex and diverse scenarios, including those it may never have encountered during training.”

Wayve shares a similar philosophy with autonomous trucking startup Waabi, which is also pursuing an end-to-end learning system. Both companies emphasize the importance of scaling data-driven AI models that can generalize across different driving environments and rely on generative AI simulators to test and train their technology.