Waymo is the first and only company operating a fully autonomous ride-hailing service to the public in the U.S. Over the years, we have been leaning more and more heavily on machine learning for all parts of our stack. As we continue to scale and advance our Driver, we want to get more out of the miles that we drive.
In this talk, Drago Anguelov, Head of Research at Waymo, will describe how Waymo leverages data collected from our fleet to continuously improve and refine our system performance. Drago will cover some recent work at Waymo on developing performant perception and prediction models using more unlabeled and less labeled data. He will also outline the role simulation can play in training robust and scalable planning systems, and describe some ongoing research on using machine learning to enhance the simulator realism, and on using simulator data to learn robust driving agent policies.MM/DD/YYYY