ADAS Meets Anthropology

Melissa Cefkin, principal scientist and design anthropologist at the Nissan Research Center, sat down with Semiconductor Engineering to talk about how people will interact with autonomous vehicles and AI and why different disciplines are required to make this work. What follows are excerpts of that conversation.

SE: Why did Nissan hire an anthropologist?

Cefkin: Anthropologists have been working in the tech sector for a few decades. By understanding people in their natural environment, and how they interact with technology products and services, you can get a deeper understanding of where new opportunities are and how to build the products and services in ways that ideally can have a long-lasting positive impact. The person who opened our lab, Maarten Sierhuis, came from NASA. He had a long history of working with social scientists, and especially anthropologist and sociologists, to help bring that perspective into technology design and use. Cars are profoundly socio-cultural objects that we encounter and deal with on an everyday basis.

SE: People are interacting with machines differently today, particularly with AI, right?

Cefkin: That’s a lot of it. And anthropologists and sociologists can squeeze out bias in algorithms. It’s not just a matter of someone having a cognitive or social response to something. It’s that systemically society produces and reproduces things. There is a much longer participation by psychologists and cognitive scientists in business settings. They are looking more into individual, cognitive, neural processes—emotional responses. Anthropologists are looking at broader aspects.

SE: So what has your research shown?

Cefkin: The place that we started was trying to understand how people react on the road in traffic with other road users so that we could begin to extrapolate what happens when you have robots in that mix. What will that be like? And what can we anticipate will be the nature of the interactions? We went through a category of interaction that we expected to be problematic, which is how people negotiate stops at intersections. We thought this would be a little bit challenging from an autonomous vehicle standpoint with turns and stops. Even though there are rules, people break the rules in a similar manner all the time. We learned that moving down the road requires micro-interactions. People need to negotiate very subtly with other people on the road, whether they are bicyclists or pedestrians or other drivers. And we do this without giving it a great deal of attention. But you can begin to see some patterns emerge as to what’s really expected. Anthropology operates from a comparative method approach. Those patterns and expectations can have slight variations from place to place. There are conventions we take for granted today, where we drive on the right side and different countries drive on the left side, and where there are lines in the road and you’re supposed to stop before those lines. Cars have a greater priority in certain locations. The things we take for granted are actually historically and socially created inventions that emerge over time, and they also change over time.

SE: There are definitely regional differences in how people behave behind the wheel. How do you design cars to deal with these variations?

Cefkin: While there are variations, some things are universal. When you travel, you can rent a car and navigation does just fine. It’s not like you’re completely hampered and unable to move. Even if you don’t rent a car, when you go out on the street you can pretty much figure out what to do. It might be a little bit more aggressive in one place or calmer. The rules of the road and the local expectations are processed just by observation in ways we can respond accordingly. Pedestrians might not wait for traffic, and if I see others walking out I can decide whether to join them. But at least I can recognize that. We are trying to be very aware of those variations of expectations, because we expect our cars to behave in ways that are comfortable and recognizable from setting to setting. On the other hand, it’s not an all-or-nothing proposition. There are a lot of different kinds of behaviors on the road. The way people expect a truck to move is very different from a bicycle, which is very different from sports car or an SUV. People set certain expectations and images. We want to learn from the core variations and respond accordingly, but I don’t think it’s an infinite variety. You also expect that people around the vehicle will adapt.

SE: But there is an element of machine bullying going on when human drivers encounter self-driving cars. People drive more aggressively sometimes around autonomous vehicles, which are pretty easy to recognize on the road. How do you deal with this aggressive behavior?

Cefkin: The fear of that problem comes up regularly. One of the ways we need to address that is to provide some very specific direction. Is there anything in the vehicle behavior that allows it to be smart enough to recognize one of those times—and for the vehicle to now assume a slightly more assertive scale? This happens now when people inch into the intersection. They’re messaging the others, ‘It’s my turn.’ That’s how they start the process, and it communicates to others that’s what’s happening. So it’s possible we can teach vehicles to take a more assertive stance. I personally think you also want to be careful before you program too much assertiveness into vehicles because this is a particular moment in history and ask how is that way of behaving on the road affect not just the cars but the whole society. And can we shape a different feeling on the road if we don’t force a ton of assertiveness and aggressiveness?

SE: So it’s not just technology that needs to change? You’re hinting at changing attitudes of people, as well, right?

Cefkin: There are people in cars trying to get somewhere, but there also are times when you’re walking down the street or riding a bike and you’re having a very different interaction. In those moments you might have a different behavior on the road. There are multiple points of view all present in different circumstances. There are many parts of the world where they are encountering autonomous vehicles with self-driving shuttles or small campuses where everyone is behaving differently. In regard to bullying, we need to be careful to not assume that people are going to take advantage of autonomous cars. People can do that today. Most cars and drivers do not want to hit people on the road and they will stop if they need to. Our fear of walking across the street when it’s not on our terms is both a fear of the vehicle and because we have social etiquette and propriety. There’s no reason to assume just because people are in a car and they’re not driving that we’re going to get in their way. Norms don’t just fly out the window.

SE: But people do behave differently behind the wheel of a car. Will there be a big change with more autonomous vehicles?

Cefkin: It’s a good question, and we don’t have good answers. People will recognize they are dealing with robotic systems, they may not recognize the actions those robots are likely to take. They’re dealing with a different entity than what they’re dealing with today. But to a large degree, it may be less significantly different than what we are already experience, with some slightly different moments of inconvenience for surprise when the vehicle around you is taking longer to speed up because that system is more conservative. This will gradualize normalize without too great of a challenge. One of the other areas we are exploring is whether we can use additional signals on vehicles to communicate what a vehicle is doing. There will be a little more information to go on. When brake lights started being used, that didn’t happen overnight. People had to understand the signals and what they meant.

Related Stories
People Vs. Self-Driving Cars
Why auto tech companies are so concerned about interactions with humans
Anatomy Of An Autonomous Vehicle Crash
Accidents happen, but with self-driving cars a crash is only the beginning.
Finding Faulty Auto Chips
The road to zero defects requires some new tactics.
The Road To Autonomy
Automotive manufacturers and chip companies reveal what they’re up to in realizing self-driving cars.