Driverless cars learn to communicate


Ford is partnering with Virginia Tech Transportation Institute to test out ways in which driverless cars can communicate their intentions to pedestrians, cyclists and other drivers.

The company points out that a simple head nod or hand wave from a driver is currently enough to indicate it’s okay for someone waiting to cross the street, but in an autonomous vehicle future, it is currently unclear how a self-driving car will make this known.

John Shutko, Ford’s human factors technical specialist, said: “Understanding how self-driving vehicles impact the world as we know it today is critical to ensuring we’re creating the right experience for tomorrow.

“We need to solve for the challenges presented by not having a human driver, so designing a way to replace the head nod or hand wave is fundamental to ensuring safe and efficient operation of self-driving vehicles in our communities.”

Ford has started the joint research project to investigate the most effective means for the vehicle to communicate.

The team considered using displayed text, but that would require people all understand the same language. The use of symbols was rejected because symbols historically have low recognition among consumers.

In the end, the researchers decided lighting signals are the most effective means for creating a visual communications protocol for self-driving vehicles. As light signals for turning and braking indication are already standardized and widely understood, they determined that additional signals could be used to communicate whether the vehicle is in autonomous drive mode, beginning to yield, or about to accelerate from a stop.

In trials, Ford outfitted a Transit Connect van with a light bar placed on the windshield. To simulate a fully self-driving experience without using an actual autonomous vehicle, the Virginia Tech Transportation Institute team developed a way to conceal the driver with a “seat suit.”

The suit creates the illusion of a fully autonomous vehicle, which is necessary to test and evaluate real-world encounters and behaviors. The researchers then experimented with three light signals to indicate the vehicle’s intent.

These were two white lights that move side to side, indicating vehicle is about to yield to a full stop; a solid white light to indicate vehicle is driving autonomously; and a rapidly blinking white light to indicate vehicle is beginning to accelerate from a stop.

The simulated autonomous Transit Connect was driven on public roads in northern Virginia throughout August, with researchers capturing video and logs of pedestrian reactions. More than 150 hours of data over approximately 1,800 miles of driving was collected in an urban environment, including encounters with pedestrians, cyclists and other drivers. External signals were activated more than 1,650 times at various locations, including at intersections, parking lots, garages, and airport roadways.

Numerous high-definition cameras mounted in the study vehicle provided a 360-degree view of surrounding areas and captured the behavior of other road users. This data will be valuable to understanding if other road users change their behaviors in response to self-driving vehicles and the signals they employ.

Ford is already working with several industry organizations to push toward creation of a standard, including the International Organization for Standardization and SAE International and it is also considering ways to communicate with those who are blind or visually impaired.