Imagine a world where autonomous vehicles are trained in a virtual realm, encountering and overcoming endless driving challenges before they even hit the road. Sounds like science fiction, right? But Toronto-based AI startup, Waabi, is making this a reality!
Waabi has created an advanced simulator, Waabi World, that trains self-driving vehicles to navigate almost any driving condition imaginable, and all in a virtual environment. This groundbreaking platform is far superior to those of competitors, as it can closely replicate real-world scenarios, including those rare, tricky situations that hardly ever happen on the road.
Founder and CEO, Raquel Urtasun, shared with Forbes that this constant learning in a detailed virtual world is preparing the software to eventually drive various vehicles, from robotaxis to semi-trucks. Urtasun, who is also a computer science professor at the University of Toronto and former chief scientist for Uber’s autonomous vehicle team, believes Waabi World is the key to unlocking self-driving technology on a large scale.
The platform is designed to be immersive and reactive. It automatically designs tests for Waabi’s self-driving brain, called the Waabi Driver, assesses its skills, and teaches it the art of driving. However, the company has no plans to license the platform to other autonomous vehicle developers, as the focus is currently on intensely training their own software.
While other leading autonomous vehicle companies, like Alphabet Inc.’s Waymo, General Motors-backed Cruise, and Amazon’s Zoox, are making progress towards broader commercialization, the pace of development has been slower than expected. These companies are performing rigorous on-road testing and simulation, but it is still unclear when robotaxis and autonomous trucks will become widespread.
Waabi, which means “she has vision” in Ojibwe, came onto the scene in June 2021, with an initial funding of $83.5 million. Although Urtasun did not disclose if additional funds have been raised since then, she did reveal that the initial focus would be on applying the Waabi Driver software to self-driving trucks, with plans to expand to all vehicle types eventually.
Urtasun pointed out that the industry’s current approach of testing and driving millions of miles in the real world is not only costly but also does not scale. Waabi World, on the other hand, provides a totally different, more efficient approach.
While Waabi seems to have a more comprehensive simulation system than currently available, Zoox’s CTO and cofounder, Jesse Levinson, noted that simulations are not perfect and real-world driving is still necessary. However, as simulations improve, there will be fewer things learned in real life that cannot be learned in a simulation.
In conclusion, while it will be some time before simulations can completely replace real-world testing, Waabi World is a game-changer. It offers a more efficient, cost-effective, and scalable solution to training autonomous vehicles, bringing us one step closer to a future where self-driving vehicles are the norm.
