Reading The Road: How Will Driverless Cars Talk To Pedestrians? – Forbes

Posted: Wednesday, January 04, 2017

No matter what vehicle we might be in, once we enter the road network, we start a constant information exchange with our fellow road-users. Signals, brake lights, and road position give important clues to those around us on what we’re likely to do next. But it’s the more subtle human-to-human interactions that are particularly important for cyclists and pedestrians – a wave of a hand, or direct eye contact with a driver, can offer a sense of road safety that technology simply can’t. So, in a future filled with free-flowing, non-traffic-jam-inducing, driverless cars, will we be able to safely cross the road?

In technology terms, there’s no reason why not. Ok yes, there’s a limit to how good a vehicle’s cameras can be – changing weather or a busy pavement can make it difficult for human drivers to spot crossing pedestrians. For an autonomous vehicle that relies on computers – even clever ones – it’s a much greater task. And then there’s the infrastructure. Even if driverless vehicles won’t necessarily need the red-amber-green signals that manage traffic today (as I described in a previous post), cyclists will, and our roads will still need to include dedicated crossing points for pedestrians. What happens at that crossing point is less clear though, and it leads onto the question of how manufacturers can build public trust in these vehicles.

Uber ‘self-driving’ cars are currently being tested in San Franscisco and Pittsburgh (Photo by Jeff Swensen/Getty Images)

Take the recent launch of Uber’s ‘self-driving’ service, for example. Video footage has shown their cars running red lights, and witnesses (including Brian Wiedenmeier from the SF Bicycle Coalition) have reported them cutting across cycle lanes – executing the so-called ‘right-hook turn’ that has previously been linked to the deaths of numerous cyclists. Not a good start. Uber also came to blows with the California DMV because they lacked the permits required to undertake these tests on public roads. Their cars are likely to return to testing in the coming weeks, but the bad press – and negative feeling amongst fellow road-users – doesn’t look like it’ll disappear anytime soon.

In 2016, Prof Natasha Merat from the University of Leeds surveyed 644 cyclists and pedestrians of all ages, in order to identify what concerned them most about interacting with driverless cars. As reported in the Guardian, the biggest request was for a clear, unambiguous way to signal to the road-user that they had been ‘seen’ by the car’s sensors – a gesture that gives them the confidence to know what the autonomous car is about to do. So what form could this sort of interaction take?

In 2015, Nissan unveiled their “intention indicator” concept, which would use flashing lights and text projected on the windscreen to communicate with humans outside the vehicle. A 2015 patent from Google shows that they’re also considering text-based displays, as well as dedicated speakers to emit audible alerts. So, there’s no shortage of ideas, but as far as I could find, few have actually been implemented and tested by manufacturers. The team at the Humans and Autonomy Lab at Duke University are a different story, though. They recently published a paper (due to be presented next week at the Transportation Research Board) which evaluated a number of display-based options for vehicle-to-pedestrian communication. (CONTINUED ON PAGE 2…)

Miles Aubert_Duke Robotics

Duke University are trying to understand how pedestrains might interact with driverless cars (Photo by Miles Aubert/Duke Robotics)


Write a Reply or Comment:

Your email address will not be published.*