It’s exciting to talk about robots as if they are the ‘human’ machines we see in movies. However, in a world striving for efficiency, robots already play a significant role in achieving this. And in a way, we are already using robots in our daily lives – think about washing machines, electric cars, and dishwashers. Or, if you look at the mass production of products and foods with the help of robots, our heavy reliance on these ‘helping hands’ is clearly visible.
Why study robot-human interaction?
As technology evolves ever so quickly, in the near future, robots could be used for an even greater variety of day-to-day purposes and services. For example, for food delivery or human transportation, as well as a city maintenance or stewardship. Or think of helping pedestrians find their way.
So, in theory, robots can be implemented for all kinds of functionalities to make the lives of humans easier or more efficient. However, at this moment, we are not yet used to robots moving around in our urban environments. How can robots support our daily lives and create added value to society? Let’s sit down for an interview with our ‘Husky’ researcher Jered Vroon.
By Yemoï Hoeben – Communications Intern at AMS Institute
Meet the Husky, our robot platform
Jered explains that the basic technical challenge of having robots moving around in our urban environments is to have them do this autonomously; finding your way, doing this safely, also by following the rules. At AMS Institute, Jered and his colleagues investigate how we can shape interactions between pedestrians and robots in cities. As part of this project, the TU Delft researcher tests the mobile robot platform ‘Husky’ at Marineterrein Amsterdam Living Lab (MALL).
With this robot platform driving around the inner-city test area of Marineterrein, the project team can evoke and understand reactions of city dwellers to autonomous robots. The social robot-human interactions the researchers are investigating, add another “raw & uncontrolled” layer to the initial technical challenge.
“Tests like these normally take place in labs or protected environments. The great thing about doing this experiment at Marineterrein Amsterdam is that the living lab allows us to conduct our research in a real-life city environment.”
Jered Vroon
Social Interaction Dynamics
A feeling of connection between humans and robots
Researching robot-human interaction can be quite complex. Among others, robots need to be designed and programmed to respond to human behavior. And as human behavior varies per person and as such is highly unpredictable, this is quite a challenge.
The team is still in the research phase at the moment, and the first question they focus on is how to design social navigation on the sidewalk. Therefore, the researchers started this project with observing how humans interact with the Husky at Marineterrein Amsterdam. By observing these robot-human interactions, the team can analyze how humans notice robots and how they instinctively react to it.
What interactions does Jered see when he’s out testing with the Husky? To illustrate, some people start teasing the robot, mostly out of curiosity how the robot responds. Other observations include cyclists stopping while the robot is crossing the road. Or groups of people getting out of the way when the robot tries to pass.
“You'd expect bystanders to be shocked by the robot. In reality, the interactions are actually very natural. Most people are intrigued by the robot in a fun way. Sometimes youngsters don’t let the robot pass by, or they are being obstructive against it.”
Jered Vroon
Social Interaction Dynamics
These make for interesting findings for the researchers, Jered indicates. For now, Jered and his colleagues aim for the Husky to contribute to a feeling of connection between humans and robots by exposing Marineterrein visitors to this robot platform.
The challenge: how to design a robot that’s ‘streetwise’?
Next to creating awareness among humans of the functionalities and purpose of robots, the researchers indicate it’s also important that this goes the other way – for a robot. For example, the Husky must know when it has priority in the streets and when it doesn’t. To illustrate, when an elderly person wants to cross the street, people instinctively stop. Ideally you'd want robots to adopt our natural instincts and behavior, Jered states.
Thus, the challenge lies in how to make a robot 'streetwise' and let it 'defuse' such situations. Ultimately, the researchers work to include these behaviors in the algorithm of the Husky.
“It’s not about prioritizing the robot, instead we want to find a harmonious way of robots co-existing with humans in a city like Amsterdam. Currently I am hard at work analyzing how pedestrians adapt to one another. And this natural dynamic people have in the streets; we try to capture that in the robot.”
Jered Vroon
Social Interaction Dynamics
From the brain of the robot to the sensors: what techniques are involved?
Looking into what techniques are involved in this robot platform, Jered explains that you can divided these in three parts. First, there’s the hardware – the brain of the robot so to say. Secondly, the robot has sensors to determine directions. At TU Delft, the researchers are developing the algorithms for the sensors to recognize “attitudes towards the robot” with machine learning. This enables the Huksy to create a 3D image of humans and makes it possible to detect their attitude towards the robot. Thirdly, the robot has physical features. In this case, in terms of motoric skills, the Husky drives around in any direction and can stop at any time you like.
Currently, the team is busy combining all these techniques in a way that the robot can autonomously make the right decision, and use the appropriate techniques in the applicable situation. This way it can read and adopt to its environment.
What tasks could robots do in our city?
It’s amazing how far we can program behavior in controlled situations, Jered highlights. For example, how cars nowadays can drive autonomously with relatively few sensors. The same accounts for Roboat – a research project by MIT Senseable City Lab and AMS Institute – a vessel that can read its environment and move around the Amsterdam canals without any human directions.
What are possible applications of robots in society this team of researchers think off? Jered indicates they explore if robots could be embedded in society for delivery purposes of, for example, packages, groceries, medications. He explains that this kind of last-mile or flash delivery could possibly be more sustainable, contactless, flexible (24/7) with robots. Or what about collection of (chemical) waste and cleaning and maintenance tasks?
“I imagine that robots can take over activities for which they are better suited than humans or that they can support humans more often with work that is, for example, very dull, dirty or dangerous.”
Jered Vroon
Social Interaction Dynamics
The future of robots in society
Jered thinks that seeing robots in our urban environments will be a process that takes years and the concept of having robots in our society has to grow. First, this study helps to understand how robots can interact with humans on a natural level. After, the functions they can fulfill can be determined. The latter will be a more design task because this has a lot to do with esthetics, he indicates.
Ultimately, Jered and his fellow researchers are excited about the future. He concludes that robots can be very capable to be part of the urban environment. In the end it is all about getting used to new situations, for both humans as robots.
