Blacksburg, Virginia - Today, a simple head nod or hand wave from a driver is usually enough to indicate it’s OK for a pedestrian to cross the street, but in a future where cars drive themselves, how will they communicate with pedestrian, cyclists or the occasional old-school human driver?
Good question - so Ford and Virginia Tech Transportation Institute came up with a user experience study to test ways for a self-driving vehicle to signal its intent, by watching the real-world reactions from people on public roads.
The researchers considered using displayed text - like the destination boards on buses - but not everybody can read the same language. Symbols were rejected for the same reason; consumer research shows they have a low recognition among shoppers.
Universally understood
But light signals for braking and turning are already standardised and universally understood, so a new set of light signals looked like the best basis for visual communication protocol for a self-driving vehicle to signal to humans (and other autonomous vehicles!) whether it is operating in autonomous drive mode, beginning to yield or about to accelerate from a stop.
So Ford fitted a light bar in the windshield of a Transit van, and six high-definition cameras to record the behaviour of other road users. Then it built a seat suit to conceal the driver, so the Transit would look like it was driving itself - that’s important to get real-world reactions between the van and other road users.
The team tried three light signals:
Yielding:
Two white lights moving side to side, indicating that the van is about to yield to a full stop.
Active autonomous driving:
Solid white light to indicate the van is driving itself.
Pulling away:
Rapidly blinking white light that it’s beginning to accelerate from a stop.
The van and its concealed driver simulated autonomous driving on public roads in northern Virginia throughout August, recording more than 150 hours of data over about 2800km of driving in urban environments, including encounters with pedestrians, cyclists and other drivers.
The signals were used more than 1650 times in and around Arlington, including at intersections, car parks, garages and the airport. Now researchers will use this data to understand how other road users changed their behaviour in response to the signals.
Project director Andy Schaudt said “This research is important not only to vehicle users and manufacturers, but to anybody who walks, rides or drives alongside an autonomous vehicle in the future.”