Podcast Notes: Autonomous Driving at Waymo
Every week, Inside AI summarizes a podcast focused on an emerging AI topic. This week features Drago Anguelov — principal scientist and head of research at Waymo — who speaks with Practical AI podcast host Chris Benson and data scientist Daniel Whitenack about autonomous driving. [Note: Questions and answers have been edited for brevity and clarity].
Daniel: Drago, could you give us your background and how you got into self-driving cars and AI?
Drago says he's a machine learning researcher and engineer by nature. He was always drawn to robotics and received a Ph.D. at Stanford in machine learning, with a focus on 3D perception under computer scientist Daphne Koller.
Through Daphne, he met research professor Sebastian Thrun, who became Drago's adviser. Drago was exposed to his work in robotics starting in the early 2000s. Slowly by surely, his path led him into self-driving.
Daniel: What’s the combination of sensors that you’re working with?
A lot of modern autonomous driving companies use all three sensors: camera, LIDAR, and radar. Some also use thermal cameras, occasionally even more creative sensors like sonar.
LIDAR stands for Light Detection and Ranging. It's a laser unit that sends out a laser pulse that then gets returned from the environment. You can measure the time travel and potentially other attributes of the signal, like the intensity of an object. It gives a very accurate depth estimate to objects.
A radar also provides a depth estimate to objects. A radar has antennae that send out RF signal, and then the signals come back, but it's not quite as localized and it’s typically lower-resolution. LIDAR - especially the high-end LIDAR - can imagine the scene at very high resolutions.
Chris: What things are you most excited about in the future of self-driving cars?
Drago is excited by the product and the possibility of helping society. There is a safety aspect. About 1.3 million people die from automotive accidents and 15 million people get injured each year. A lot of this is due to human error. With autonomous driving systems, the hope is to reduce the consequences and the possibility of human errors.
Chris: Can you envision a world where humans are not driving at all anymore?
It may take a while to get there. People are attached to their cars and penetration will happen in a thoughtful, steady way over time. There is the hope that people will be attracted to the advantages of autonomous driving...