How Waabi’s Raquel Urtasun Is Building One A.I. Brain to Drive Trucks and Robotaxis

Waabi founder Raquel Urtasun is developing a single AI system that can handle both trucking and robotaxis, aiming to create a scalable, generalizable brain without relying on massive real-world data. The company trains its AI in a hyper-realistic simulator, allowing it to learn from few examples and adapt across vehicle types. This approach contrasts with traditional methods that require extensive data collection and manual coding. Urtasun believes this creates a more efficient and scalable system, as it avoids the need for separate technologies for each use case.

The AI system is being deployed on freight trucks in Texas, with plans to go fully driverless next year. Waabi has also partnered with Uber to bring its technology to 25,000 robotaxis. This represents a significant shift from earlier autonomous driving approaches, which relied on manual coding and large data sets. Urtasun argues that the new method allows the AI to generalize across different environments and vehicle types, reducing the need for specialized systems.

The AI system is designed to handle complex scenarios, such as driving large trucks with heavy weight distribution, which are difficult for humans and machines. Waabi's approach allows the AI to learn from limited examples and apply its knowledge to different platforms. This is different from traditional black-box systems that require billions of miles of data to train. Waabi's technology is seen as more efficient and safer, as it can be validated in a simulator before being deployed on real roads. The company is waiting for Volvo to complete validation before launching fully driverless trucks, emphasizing safety and regulatory compliance.

AI-generated summary of the original article — see the source for the full story.