One of the mysteries related to Uber’s sale of its Uber ATG self-driving unit to Aurora has been solved.
Uber ATG’s chief scientist, AI pioneer Raquel Urtsun, has launched a new startup called Wabi that it describes as an “AI-first approach” to accelerate the commercial deployment of autonomous vehicles, starting with the long- haul trucks. Urtasan, who is the sole founder and CEO, already has a long list of high-profile backers, including separate investments from Uber and Aurora. Wabi won a Series A round led by Khosla Ventures with additional participation from Uber, 8VC, Radical Ventures, Omers Ventures, BDC, Aurora Innovation as well as leading AI researchers Geoffrey Hinton, Fei-Fei Lee, Peter Abeel, Sanja 83.5 million dollars raised. Fiddler and others.
Urtasan described Wabi, which currently employs 40 people and operates in Toronto and California, as the culmination of her life’s work to bring commercially viable self-driving technology to society. The company name – wabi means “he has vision” in Ojibwe and “simple” in Japanese – hints at his vision and ambitions.
Autonomous vehicle startups that exist today use a combination of artificial intelligence algorithms and sensors to handle the tasks of driving that humans do such as detecting and understanding objects and making decisions based on that information on a deserted road or road. To navigate the congested highway safely. Beyond those basics there are a variety of approaches within AI.
Most self-driving vehicle developers use a traditional form of AI. However, the traditional approach limits the power of AI, Urtsun said.Developers have to manually tune the software stack, a complex and time-consuming task. Urtasan says: Autonomous vehicle development has slowed and the limited commercial deployments that exist operate in smaller and simpler operational domains because scaling up is so costly and technically challenging.
“Having worked in this field for so many years and, in particular, the industry for the past four years, it has become way more and more clear that there is a need for a new approach that differs from the traditional approach that most companies have. Taking it today,” said Urtsun, who is a professor in the Department of Computer Science at the University of Toronto and co-founder of the Vector Institute for AI.
Some developers use deep neural nets, a sophisticated form of artificial intelligence algorithms that allow computers to learn using a series of connected networks to identify patterns in data. However, developers usually close the deep net to handle a specific problem and use machine learning and rule-based algorithms to tie into the wider system.
Deep nets have their problems. A long-standing argument is that they cannot be used with any reliability in autonomous vehicles because of the “black box” effect, in which How And this Why A specific task the AI solved is not clear. This is a problem for any self-driving startup that wants to be able to verify and validate their systems. It’s also difficult to incorporate any prior knowledge about the task the developer is trying to solve, like, uh, driving for example. Lastly, learning deep nets requires a huge amount of data.
Urtasan says she solved these complex problems by combining them with probabilistic inference and complex optimization around deep nets, which she describes as a family of algorithms. When combined, the developer can trace the decision process of the AI system and incorporate prior knowledge so that they do not need to teach the AI system everything from scratch. The final piece is a closed-loop simulator that will allow the Wabi team to test it extensively in typical driving scenarios and safety-critical edge cases.
Wabi will still have a physical fleet of vehicles to test on public roads. However, the simulator would allow the company to rely less on this form of testing. “We can also prepare for new geographies before we drive there,” Urtsun said. “That’s a huge advantage in terms of scaling curve.”
Urtasun’s vision and intention is not to adopt this approach and disrupt the ecosystem of OEMs, hardware and compute suppliers, but to become a player within it. This may explain the support of Aurora, a startup developing its own self-driving stack, which it hopes to deploy first in logistics such as long-distance trucking.
“It was a moment to really do something different,” Urtsun said. “The field needed diverse approaches to solving this and it became very clear that this was the way to go.”