Drivers have plenty of ways to communicate with their fellow drivers, beyond the basic turn signal. Some methods are nice, like a friendly wave to beckon a merging car into your lane. Others, not so much. We’ve all probably cursed or honked at another driver, or been given the finger.
But if driverless cars become a reality, those vehicles will also need to be able to reliably communicate with each other, using wireless communications technologies.
That was one of the issues raised Wednesday during a panel discussion at a conference on artificial intelligence and machine learning held at the MIT Media Lab. The event was hosted by the MIT Initiative on the Digital Economy.
“There is, in fact, a lot of effort to have robots and A.I. systems talk to each other,” said Manuela Veloso, a Carnegie Mellon University professor and head of its machine learning department. “The challenge for us is how do we communicate with another robot that was not developed by us.” For self-driving car developers, that means “how is a car from Toyota going to talk to a car from Uber or Google?”
It’s an important question, and one that seems worth asking now, given the already fierce competition among car manufacturers and tech companies developing autonomous vehicles. But when it comes to enabling vehicle-to-vehicle communications, the sector’s various players are apparently willing to collaborate and share information.
“In order for it to work, it will have to work across brands,” said Gill Pratt, the CEO of Toyota Research Institute. “It’s not that we share everything. But we try as much as we can to share things that are related to safety and to traffic.”
Companies are working together to develop technical standards for wireless communications between vehicles, Pratt said. They’re also trying to ensure some of the wireless spectrum—which is regulated and licensed by the Federal Communications Commission—is reserved for vehicle communications, he said.
Meanwhile, in December, the U.S. Department of Transportation proposed a rule that would require vehicle-to-vehicle communication capabilities in new cars, The Verge reported. The department believes the technology could help human drivers (and eventually autonomous vehicles) avoid accidents.
For autonomous vehicles, vehicle-to-vehicle communication “matters a lot because it can drastically improve the safety of these cars working together,” Pratt said. “It helps if my car can see what it sees, but sees what the other cars can see, too.”
There’s also work being done to enable vehicles to communicate with other types of machines. There are experiments being run in Japan, for example, that install cameras above intersections at autonomous vehicle test sites, Pratt said. These cameras broadcast information about the view from above to driverless cars in the vicinity. They might be able to help the vehicles “see” around the corner if it’s blocked by a building or other traffic, he said.
Of course, even if companies develop reliably safe autonomous vehicles, roads won’t transition from all human drivers to all machine-controlled vehicles overnight. That means driverless cars will need to be able to communicate with human drivers and understand their behaviors.
“The disruption yet to come is … learning systems that predict how humans behave,” said John Leonard, an MIT professor who researches navigation and mapping capabilities for autonomous mobile robots.
After the panel discussion, Pratt told Xconomy that Toyota Research Institute officials are thinking about ways to enable autonomous vehicle systems to interpret human communication signals, such as hand gestures. But those efforts are still at the research stage, he said.
“The trouble is, it’s really hard,” Pratt said. For example, a hand wave from a driver sometimes means “go” and sometimes means “stop,” and the differences are subtle.
Another issue is that autonomous vehicles will need to be able to recognize signals from a police officer directing traffic in the street—and know to obey that person. One idea, Pratt said, might be for the officer to have a device that wirelessly communicates commands to an autonomous vehicle.
“I think there are ways to make it easier for machines by having people adapt,” Pratt said. “We don’t have the answer yet.”