Self-driving cars are all over the news these days, especially here in Austin – the second city in which Google has rolled out the new technology. From befuddling track stands and gratuitously courteous technology to potentially big gains for safety, self-driving cars are already changing the urban fabric of our city. What all this means for bicycling in Austin is of course top of mind for Bike Austin, so we thought we’d go straight to the source. Google’s Head of Business Operations for the Google Self-Driving Car Project Jennifer Haroon fielded some of our questions:
What do you see as the biggest benefits of self-driving cars for pedestrians and bicyclists? What about potential dangers/drawbacks?
We’re hoping to improve road safety for all of us. Self-driving cars never get sleepy or distracted like humans too often do when they’re behind the wheel. Our sensors can see 360 degrees out to nearly two football fields in all directions and simultaneously detect many objects around them. We think this could help avoid lots of sticky situations on the road.
One of the challenges we face in developing our software is that humans can be pretty unpredictable! Fortunately, over many miles of driving experience, we can even learn to “predict the unpredictability” and navigate safely and appropriately. And if the car isn’t sure exactly what to do, it will do the safe thing and slow down or come to a stop.
What is google doing to ensure that self-driving cars recognize and avoid confrontations with bicyclists on the road?
Our software classifies cyclists distinctly from other vehicles or pedestrians, so we know generally what to expect from them when we detect them on the road. For example, we can detect hand signals that indicate a turn, and we can detect behavior like wobbling or weaving that might indicate they’re about to change direction. We’ve also taught the car to nudge away from cyclists to give them a little more room.
Bike Austin strongly supports protected bike lanes for bicyclists – do you see self-driving cars and any infrastructure that might accompany them peacefully co-existing with bike lanes in the future?
Absolutely. Our highest priority for the project is improving safety on the roads. If we can help make cars safer, that will hopefully make cycling safer too.
We’ve read about the fixed-gear bike tricking the self-driving car in Austin – What have been some of the biggest challenges in integrating self-driving cars into a complex urban landscape with bikes, pedestrians, etc?
As it turns out, what looks chaotic and random on a city street to the human eye is actually fairly predictable to a computer. As we’ve encountered thousands of different situations, we’ve built software models of what to expect, from the likely (a cyclist riding with the flow of traffic in a bike lane) to the less likely (riding straight at us, directly in our lane…it has happened!). We continue to drive thousands of miles every week to gain more experience and uncover new challenges; the track stand is a great example of why that’s important. When we do have an interesting encounter, our test drivers make note of it and pass the feedback to our engineers, so we can refine our software. It’s also great to be in Austin, where we can see how road behavior and people’s response to our cars might be different than in Mountain View, where people have been seeing us around for years.
Anything else the bicycling world in Austin should know?
Please tell us how we’re driving! In this phase of our project, we’re focused on learning how people in the community perceive and interact with self-driving vehicles. For more information on how our cars work, as well as a feedback form where you can let us know your thoughts, please visit http://www.google.com/