Wayve, an autonomous vehicle startup, is focusing on the potential of its technology in the market, according to co-founder and CEO Alex Kendall. The company aims to keep its automated driving software affordable, adaptable to various hardware, and suitable for advanced driver assistance systems (ADAS), robotaxis, and robotics.
During Nvidia’s GTC conference, Kendall outlined Wayve’s strategy, which involves an end-to-end data-driven learning approach. This system interprets input from various sensors, such as cameras, into driving actions like braking or turning. Unlike previous autonomous vehicle technologies, Wayve’s system does not depend on high-definition maps or rules-based software.
Wayve, which began operations in 2017, has managed to attract investors, raising over $1.3 billion in the last two years. The company plans to license its self-driving software to automotive and fleet partners such as Uber. While it has not announced any automotive partnerships, a spokesperson revealed that Wayve is in advanced discussions with several original equipment manufacturers (OEMs) to incorporate its software into different vehicle types. A key selling point for Wayve is its cost-effective software, which does not require additional hardware investment to integrate with existing sensors in vehicles.
The company’s approach is “silicon-agnostic,” meaning the software can operate on any GPU that OEM partners already use, though Wayve’s current development fleet utilizes Nvidia’s Orin system-on-a-chip. Entering the ADAS market is crucial for Wayve, as it enables the development of a scalable business model, broader distribution, and data exposure necessary for advancing the system to Level 4 autonomy, which allows the vehicle to navigate independently under certain conditions.
Initially, Wayve plans to commercialize its system at the ADAS level, which means the AI driver is designed to function without lidar, a sensor most companies developing Level 4 technology consider critical. Wayve adopts a similar autonomy approach to Tesla, leveraging end-to-end deep learning models and ADAS deployments to refine its self-driving software. However, unlike Tesla, which solely relies on cameras, Wayve is open to using lidar to achieve near-term autonomy.
Alex Kendall mentioned the potential for future sensor suite reduction, depending on the desired product experience. For instance, additional sensors may be required for driving in conditions like fog. He also discussed GAIA-2, Wayve’s latest generative world model for autonomous driving, which trains its AI driver on extensive real-world and synthetic data, allowing it to exhibit more adaptable and human-like driving behavior.
Wayve aligns its philosophy with that of the autonomous trucking startup Waabi, both of which pursue an end-to-end learning system, emphasizing scalable data-driven AI models usable across diverse environments. Both companies utilize generative AI simulators for testing and training their technologies.
