On Teaching Autonomous Vehicles to Drive
Last week, Dmitri Dolgov, Co-CEO of Waymo (an autonomous vehicle spinout from Google), said in the blog post “How we’ve built the World’s Most Experienced Urban Driver”:
“We’re building systems that operate without any reliance on human drivers … we focus on the hardest parts of the challenge, which is critical to reliably scaling vehicles with no human drivers.
“With more than 20 million miles autonomously driven on public roads and over 20 billion miles in simulation under its belt, the Waymo Driver develops an incredibly nuanced understanding of a city’s roads and its drivers’ behavior to adapt to the local driving character … we’ve expanded our testing in San Francisco, where we’re currently driving more than 100,000 miles per week.
“The fifth-generation Waymo Driver is our most capable and advanced system yet ... comprised of complementary sensors, including radar, lidar, and cameras, it can see 360 degrees around the vehicle, day and night, even in tough weather conditions such as rain or fog.
“Our powerful onboard compute platform allows the Waymo Driver to process vast amounts of data and run real-time inference on large Machine Learning (ML) models to react immediately without additional human input, like when an emergency vehicle is approaching, or when a dog runs out into the street.”
Separately, the Rand Corp. report “How Many Miles of Driving Would It Take to Demonstrate Autonomous Vehicle Reliability” (2016) said:
“Autonomous vehicles would have to be driven hundreds of millions of miles and sometimes hundreds of billions of miles to demonstrate their reliability in terms of fatalities and injuries … test-driving alone cannot provide sufficient evidence for demonstrating autonomous vehicle safety.
“Developers of this technology and third-party testers will need to develop innovative methods of demonstrating safety and reliability …. Even with these methods, it may not be possible to establish with certainty the safety of autonomous vehicles. Uncertainty will remain.”
Table: Examples of Miles and Years Needed to Demonstrate Autonomous Vehicle Reliability
Source: Rand Corp.
OUR TAKE
Expected benefits from AV use include 1) improving safety (over 90% of crashes in the US result from human error), 2) increasing access to elderly and disabled individuals, and 3) lower the costs of commercial transportation services.
Issues to address include 1) determining liability when an accident occurs, 2) the displacement of commercial drivers, and 3) data privacy related to AV operations.
While there is uncertainty about the number of training miles needed to make AVs safe (and how data will adapt to new technologies and new vehicle models), San Francisco residents know that Waymo vehicles are on the road all day collecting data to help reduce the uncertainty.