Driverless cars getting into fatal accidents are raising serious doubts over the safety of these vehicles. But more important than that is how people interact with these vehicles and how these vehicles interact with society. In a traditional setting, when a car approaches a crowded street, the driver can use subtle signals to tell the pedestrian whether to move first or suggest that he be allowed to move first himself by slowly moving the car. Similarly, the pedestrian can signal a car to stop and allow him to cross even if it's a green signal for the car.

Similar interactions would be challenging in an environment where there are several autonomous cars and even among autonomous cars there are some being driven by a physical driver and some without. Several companies are trying to work on this problem to make autonomous cars more socially acceptable.

Nissan is investing heavily not just in driverless technology, but also in how to make this technology interact better with humans. "My goal at Nissan and doing this kind of work is that if we don't take a large step forward ourselves, trying to make our cars move about in harmonious ways with society, then not only will we have a big acceptance problem on our hands, we would also have a greater likelihood of creating secondary outcomes and problems that we might not even anticipate," Melissa Cefkin, Principal scientist and design anthropologist, Nissan Research Center, told BusinessLine .

"We are studying what difference it would make if we add some additional signalling outside autonomous cars. In particular, we are focusing on the intent problem. So, how a robot car, which in the future might not have a driver to help it directly or indirectly send a signal to a pedestrian or another driver. Once we have those kinds of minute forms of human interactions built into the system, then maybe we will benefit from using different kinds of communication systems outside so that an autonomous vehicle can communicate with others in the environment about what it's doing or what it's about to do," Cefkin said.

But training the cars may not be enough and some amount of training would also be needed to be given to people on the streets, feels Cefkin. "Inevitably, people will need to learn in a small nuanced way that they are interacting with something that doesn't move quite like them. Because the vehicles are being designed for safety or overall efficiency and safety on the road, they'll make decisions in a more predictable way. But what is predictable in one setting would be quite different in another. But because they will operate differently, people will need to adjust how they interact with autonomous cars."

Testing these interactions are also not as easy as they sound. For one, people behave differently when they see a marked autonomous car for the first time versus when they seeing it for a 100th time. Moreover, with so many autonomous cars on the roads in future, being able to differentiate them from those being driven by a real driver could be a challenge and could lead to accidents. Nissan, though, is trying to create every possible test scenario to make autonomous cars as acceptable to society as possible.

comment COMMENT NOW