It’s easy for us to forget that the vast majority of the world doesn’t really care about (or even know about) robots. With that in mind, it’s understandable why most roboticists consider robots operating “in the wild” to be “anywhere that isn’t the controlled environment of my lab.” But there are “real world” environments, and then there’s the actual wild, and we almost never hear about research happening there.
This is too bad, because we don’t have nearly enough appreciation for how robots can potentially be used to mitigate problems throughout the developing world. There’s also very little research into how different cultures react to robots with a social component—most human-robot interaction (HRI) studies rely on local participants who are easy (and cheap) to recruit, and are consequently full of students, which is a terrible representation of most of the rest of the world.
At the Social Robots in the Wild workshop at HRI 2018, Dr. Amol Deshmukh from the University of Glasgow (along with collaborators from AMMACHI Labs at Amrita University in India) presented a new study where they took a water hauling robot to a village in rural India. The wild here isn’t just about remoteness and lack of infrastructure—it’s also about the fact that none of the people in this study had ever seen a real robot of any kind before.
Water transport is a real problem in rural India. Over half of the population doesn’t have tap water at home, so people (usually women) have to go get water from a central location and haul it back. They spend roughly 5 hours every day doing this, carrying jugs of water on their heads weighing up to 20 kilograms. It’s not good for their bodies, and it’s not good for their schedules, taking time away from other activities that are almost universally more productive (if less thirst quenching).
About 200 people live in the village of Ayyampathy in the southern Indian state of Tamil Nadu. The 25 houses in the village are between 100 and 500 meters from a nearby water tank, and households make an average of 15 water carrying trips per day. Moving 20 kg of water at a time in jugs is no fun at all, especially in the summer heat. You’d think that if someone showed up with a water hauling robot and offered to help, it would be an easy sell, but in practice, things are much more complicated than that, as the researchers explain:
Most subjects lived on a daily wage basis and had their own daily routine to carry out their household chores like cattle feeding, cooking, washing etc. They were not keen to commit a fixed time for the study. It was difficult to predict when they would need to fetch water. They had different water requirements and time schedule for fetching water. Some would do it early in the morning, others later during the day. It was challenging for researchers to be present with the robot when they required water.
Approaching people to volunteer for the study was particularly challenging due to cultural constraints. The women in the village were hesitant to communicate initially with male researchers in our team. We had to recruit a female researcher to ease the flow of communication.
The researchers ended up with 11 participants willing to try to work a robot into their water carrying routines, including 10 women and one man, ranging in age from 15 to 70. The gender imbalance is due to the fact that women are usually the ones hauling water. None of the participants had ever seen a real robot before, although two of them had previously seen robots on television, and these two (who had high school education) were also the only participants who had been exposed to computers.
Clearpath Robotics’ Husky robot.Photo: University of Glasgow & Amrita University
The experiment itself was straightforward: a Clearpath RoboticsHusky was equipped with a crate that could hold three jugs of water, totalling 60 liters, much more than a single person could carry by themselves. A Bluetooth speaker allowed the robot to communicate with people in a synthetic male voice (speaking Tamil, the local language). The robot wasn’t autonomous; a researcher remote controlled it from about 10 meters away, and this method of operation was made clear to the participants in the study.
When it was time to get water, the robot would wait its turn, and then the study participant would fill each jug and load it up. The robot would then politely ask, “Can you show me the way to your home so that I can bring water to your home?” Once there, the participant unloaded, emptied, and replaced the jugs, and the robot would say, “I hope I was helpful to you. Please remember to wash your hands before you eat.” How nice.
A follow-up interview asked their participants about their experiences with the robot, and their answers illustrate why this kind of HRI research in the wild is so important. An example from the paper:
One hundred percent of the participants perceived the robot as being alive, attributing aliveness to the movement and speech of the robot, for example two of them said “Without being alive, how can it talk?”, “Only because it has life, it is following us, right?”
At the same time, the participants were fully aware that there was a human controlling the robot, and most of them expressed some amount of confidence that they’d be willing to operate it if given lessons. Generally, the impressions of the robot were positive—everyone said that it made the water hauling job easier, and most felt like the experience was safe and enjoyable.
Despite the fact that the robot had a male voice, four of the 11 participants described it as a female. Three didn’t know what gender it was, two said it didn’t have a gender, and only one (a woman) said she thought the robot was male. It’s remarkable that the gender bias here is entirely centered on the task that the robot is doing, overcoming even an obviously male voice.
As the researchers quite rightly point out, “Most HRI research is carried out in urban environments with people from developed countries. HRI in rural settings is a hugely unexplored research area where social robots could potentially be used to create a positive impact for the wider community rather than a privileged few in the developed world.” But in order to achieve that positive impact, we have to understand the cultural context, and that means finding ways to do more research like this. In the wild. For real.
Photo: University of Glasgow & Amrita University
For more details on this study, we spoke with lead author Dr. Amol Deshmukh.
IEEE Spectrum: It seemed like you had a little difficulty at first convincing people to try the robot out, is that right?
Dr. Amol Deshmukh: Initially, they absolutely didn’t know what to expect. They were very apprehensive. Once we got the first participant on board, and they saw what was happening, all of them jumped in and said, “Okay we’d like to use the robot as well.”
The other thing which was very interesting for us was that the first day was a failure. The problem we had recruiting participants was that almost all the water carriers are females, and it was very difficult to get them communicating with the male researchers on our team, because of the cultural differences. We went back and recruited a female researchers who accompanied us the second day, and it was easier for her to communicate what we were trying to do. These kinds of cultural differences matter when you’re trying to run studies like these in rural areas.
With this research, I wanted to make a point to the HRI community, in a way, that they should look at these subjects as well—where potentially the application can have a huge influence and impact in the daily lives [of people in developing communities].
There was a researcher following behind the robot and operating it; how did the villagers understand the relationship between the robot and the person controlling it?
The villagers understood that the robot was being operated [by a researcher], but in a real sense, it didn’t matter. The social perception of the robot was as a separate, autonomous entity. It would be interesting to do this in a more truly autonomous way, but it would be a very challenging thing to achieve in a cluttered outdoor environment.
Would you expect there to be more of a focus on the person driving the robot if you were to repeat this study in a more urban environment? Would the idea of autonomy be fundamentally different?
I believe that the perception would be different. That’s the main novelty of this study: These people are hardly exposed to technology, they don’t know anything about technology. In urban environments, people are exposed to technology, to science fiction, and they build their own mental models about robots. But exposing people who are on the opposite end of the spectrum—that was the main focus of this research.
Why do you think there hasn’t been much of a focus on research like this?
I think there are two reasons. One is the commercial aspect—there is hardly any money there. Most of the robotics work, HRI work, or social robotics companies are targeting urban populations as consumers who can buy their robots. These rural subjects are not thought of as ideal customers because nobody can afford this technology. This is common with technology in general: you see all the cool stuff that’s being developed, and I don’t think anybody is thinking about this subject pool, and it’s a massive population. Half of the population on the planet is not privileged enough to have education and access to technology.
The other reason is of course logistics. There are hardly any houses with electricity, so how would they charge a robot? It’s a kind of strange example where these people don’t have access to basic things and we’re providing them with a robot to carry water. It’s a mismatch, but that’s what has made this interesting.
Photo: University of Glasgow & Amrita University
Does it make sense to try to leapfrog basic development like plumbing or an electrical grid and deploy a robot to carry water?
No. Of course they need the basic amenities—they need their houses to be connected with pipes, they should have access to tap water. That is one thing that has to be a priority.
So in what sense could a system like this be practical?
There are some villages in the desert areas of India where people have to walk 5 to 10 kilometers every day during summer [to get water], in extreme heat. I think there is a lot of need there to provide technological interventions; the government is trying its best to provide water for them, but it’s not easy, so I think there is a lot of potential there for this technology to be impactful by carrying water for them.
What are you working on next?
I had a little bit of funding to go to India to do this research; I’m looking for some funding that is more substantial so that we can take this to the next level.
Right now, I’m actually working on an EU project with a Pepper robot, deploying in a shopping mall and doing research on that. It has been a question that has bothered me for years now—is it really causing impact? All these HRI papers, are they actually causing real impact to people’s lives? So I came up with this idea exploring this small [India] project, and the feedback has been really positive. We’re exploring funding options, and if we get more funds, we could roll out more of these kinds of robots and carry out the research in a more extensive way.
“HRI ‘In the wild’ In Rural India, A Feasibility Study,” by Amol Deshmukh, Sooraj Krishna, Nagarajan Akshay, Vennila Vilvanathan, Sivaprasad J. V., and Rao R. Bhavani from the University of Glasgow and AMMACHI Labs at Amrita University, in Kerala, India, was presented this week at the Social Robots in the Wild workshop at HRI 2018 in Chicago. This research was funded by Scottish Informatics and Computer Science Alliance (SICSA) and Ammachilabs, Amrita University, Kerala, India.
Evan Ackerman is a senior editor at IEEE Spectrum. Since 2007, he has written over 6,000 articles on robotics and technology. He has a degree in Martian geology and is excellent at playing bagpipes.