Nissan Is Using Anthropology To Teach Autonomous Cars How To Behave
It may be technically easy to make a car drive on its own, but eventually they have to deal with complex social interactions like hand signals, head nods, horns, or even road rage. We humans, as their ultimate end users, and fellow road users, are too diverse, complex, and unpredictable in our behaviour and that’s true with our driving habits as well. Dr. Melissa Cefkin, a design anthropologist working at the Nissan Technical Center in Silicon Valley, is given with a ‘simple’ task – teaching self-driving cars how to behave responsively to the human interactions on roads through novel semaphore systems.
Dr. Cefkin’s is in charge of the research into the complex web of human-machine interaction, within and outside the vehicle, and in the context of the broader culture. He prime research questions remains to be “how many regularities and universalities are there in human behavior, or is there just kind of infinite variation?”
The project takes an empirical look at what happens in shared spaces (such as roads) between and among humans and vehicles, so as to decode certain patterns of interaction. These insights can aid the research and development of autonomous cars that are going to be a big technological breakthrough in the years to come. The responsive driving manners of robotic cars have direct impacts on two most important objectives behind these cars – reliable automation of manual driving and enhanced road safety, thereby eliminating plausible human errors committed on roads. The core tools of anthropological research such as individual and group interviews, charted observation, and codified ethnographic videography are used in order to decipher patterns of interactions that can be fed into the decision chains of self-driving cars.
Interestingly, some of their research findings apparently reflect the key challenges that autonomous vehicles face on roads. Since these cars are basically designed for efficiency and lack self-thinking as any other machines, they generally tend to be conformists to road rules. For instance, at traffic signals and regulated road intersections, robotic cars do fairly well as their human counterparts.
However at abstruse road situations like that of a typical Old Delhi streets, their driving can be confounding. Cefkin, at an interaction with Brett Berk of Drive.com, says that there are push and pull factors, varied interpretations of what stopping means and in what order it should occur or cease at ‘artifacts’ like stop sign where people have to make conscious decisions of what they want to do.
Further, she observes that there seems to be a predominant preference for maintaining some sort of movement and flow, with respect to pedestrians, skateboarders and scooterers on American roads and sidewalks, akin to Newton’s Law of Inertia. “There is a very subtle selection of pathways and routes that they use to avoid interaction requirements—places where they’re going to have to stop and figure out, okay, what do I do”, she says. As expected, machines may find it hard to predict such dynamic and enigmatic human expressions that are subtle and sudden. Bicyclists are perfect example for Cefkin. “Bicycling is a whole social movement. It’s a cause. It’s a mission—there are activists in this arena”, she adds.
In every such encounters, non-verbal signs such as head nods, hand signals, horn honks, headlight flashing, and terse expressions implying annoyance or violence exist in communications between road users. So as to make self-driving cars to imitate such means, Melissa Cefkin is assisting Nissan to come up with some sort of modern semaphore systems. Case in point are color-coded lights that might let proximate human drivers know of the car’s intent to start, stop, or stay in place. Light-eye strips may track their movement, alerting them to the vehicle’s surveillant awareness of their presence. And a text panel could flash directional messages like “After You,” and “Please Wait”.
Moreover, this kind of an anthropological approach leads us to further questions. Do automated cars bring any alternative and more physical experiences in moving through time and space, interacting with surroundings and environment? Do they add anything into the relationship between the outside and the inside, and what people might be doing and experiencing as they move through space driven by cars on their own? Liberated from the need to drive, or from the division between driver and passenger, can robotic cars help bring the outside in, or facilitate a more shared experience?
Melissa is of the view that automated cars offer a specific and unique kind of interaction, allowing us to take in and look together at the same space as we move forward through it. But it is also possible that occupants may further retreat into their individuated virtual environment created by their smartphones or tablets, as in the case of our metro trains.
“The trend of being cloistered in our own worlds and living in a bubble through our communication devices, there’s no reason to think that automatically or magically changes with autonomous vehicles,” Cefkin says. “But occasionally I hear of people getting themselves off social media intentionally, or investigating social movement niches like sustainability and the search for new collective forms of economic and social interaction,” she says positively.
After all, automakers are interested in the insights anthropologists can bring into how people are using their products and the experiences they create. This is very much true with the ‘driver-less’ cars that are fast evolving, as the term does not mean ‘people-less’ cars. A critical understanding of road interactions and driving realities that are fundamentally human and therefore ‘social’, in addition to the symbolism involved, is essentially in building a reliable self-driving car.
HT and Photo Credit: DRIVE.COM