When I began my doctoral studies, I had a dilemma. I thought: I am going to develop robots which may then take up arms and hurt people – says Krzysztof Walas*, PhD, Poznan University of Technology, in the interview conducted by Anna Zagórna
Anna Zagórna: What should robots know and be able to do in 20 or 30 years?
Krzysztof Walas, PhD*: The biggest challenge is to make them behave and act in a way that is neutral to humans and their environment. Their actions and movements should not make people afraid of their company.
This is achievable in robotics over the next decades. Research on teaching robots how people enter their personal spaces and how they protect them is already underway (although it is not publicised).
Whenever we see a robotic arm in operation, we are afraid of getting hurt or hit. We fear some gestures more than others. If someone gets too close to us and makes an odd movement, our unconditional reaction is to avoid it in fear of something bad happening to us. We are similarly influenced by the behavior of robots.
How to make such progress?
You need the robot to merge with and melt in reality. The robot of the future should possess information about our presence, blend in with the real environment so that we do not have to be afraid of it. For example, we should not run away from it or leap out of its way when we see it walking on the sidewalk, as it will know the rules of giving way to other pedestrians.
It is a lot of work and it goes beyond robotics; it also requires interdisciplinarity and very intense scientific cooperation. We must ensure that in 20 or 30 years ethical and sociological standards as well as principles of cooperation with robots are already in place. As technology is developing rapidly, I do not think that it will inhibit the process of achieving that goal. The real challenge is to integrate robots with the society.
With our help, machines are getting increasingly smarter. Is there anything to worry about?
I think so. But that does not result only from the fact that robots are getting increasingly “smarter”. When I began my doctoral studies, I had a dilemma. I thought: I am going to develop robots which may then take up arms and hurt people. It was at that point when one of my professors told me that you could use a knife to eat but also to hurt. The same goes for robots. Their behavior and course of action depend on humans. In the future, a threat will not come from robots but from humans who decide to use them for the wrong reasons.
However, in the case of robots equipped with artificial intelligence, the situation is more complex. There is the issue of the so-called technological singularity. It is the moment when the robots will surpass humans in intelligence and gain consciousness, and we will not be able anymore to keep up with them. Then, they might say that something is wrong and come to the conclusion that people destroy the Earth and therefore constitute a threat to the world.
Will we love or hate robots?
I guess I am too old, because I cannot even imagine that. I have not lived in the virtual world long enough. But I do not say it is impossible. Many science fiction films have been showing what is happening in reality for a long time. Let us take a look at Japan, for example. Some time ago young people used to play with primitive tamagotchis, which are now nothing but retro toys.
You can use a knife to eat but also to hurt somebody. The same goes for robots. Their behavior and course of action depend on humans
Nowadays, many young Japanese men spend time talking to a chatbot girl on their mobile phones because they simply like it. She is nice, she does whatever she is asked to do, she does not sulk about anything, she has no expectations, you can get emotionally attached to her. There are also youths who live in a virtual world and for whom the line between reality and virtuality is blurred. The feelings we have had for living beings can be showed to machines.
Are there any innovations in the world of robots and artificial intelligence that surprised you?
Yes, for example TIMESTORM project, which is focused on how robots perceive time. It was carried out within the scope of the Future and Emerging Technologies program of the European Commission, by, among others, researchers from the Karlsruhe Institute of Technology.
They wanted to find out how to make robots have a sense of time that would be similar to the one humans have. People do not need calendars to tell and feel how much time has passed since someone’s departure last week or how much time will pass before an event scheduled to take place in one month. That was a surprising and very bold research approach.
What would be a breakthrough invention for robotics?
Efficient power supply. The energy autonomy of mobile robots is so low that they can perform missions which are not longer than several hours. The question is how to make them load faster and how to prevent them from running out so quickly.
It seems that atomic fuel or fusion of atoms are not coming any time soon. Now we are trying to optimize performance and power. If we take a closer look at mobile robotics, we will notice two important technologies: energy density and torque density, i.e. whether it is possible to concentrate more energy in a smaller battery mass. The same goes for the engines: a higher torque generated from smaller mass means a few more minutes of operation.
Your dream is to see Poznan as the center of Polish robotics. How can you make your vision come true?
Poland needs a location that would be immediately associated with world-class robotics. As a local patriot, I would like high school graduates to immediately associate robotics with Poznan. I would also wish they came to study in my city. Everything is now going in the right direction. We have taken steps to intensify international cooperation in terms of H2020 projects and to organize pan-European events which will help us to stand out globally.
What should be done to make the quality of teaching a decisive factor?
We are working on that. For example, by organizing a big, valuable scientific events. We need to show that what we are doing here is top-quality and that the world is interested in that. To attract the world’s attention to my projects, I have to be involved in both marketing actions and scientific research.
I believe that in this way we can lure students to Poznan and that they will want to learn in our team, as we collaborate with the University of Oxford or the Massachusetts Institute of Technology. We want them to see that our university offers an opportunity to meet best scientists or to serve an internship abroad. We want them to perceive us as their window on the world, a gateway to the best international laboratories, but also as a place to send the best students for research exchanges. And it is not about brain drain. It is about benefits for both parties.
Several years ago, within the framework of the “Leader” program competition organized by the National Center for Research and Development, you received a considerable amount of money for your project called “Perception and control in the task of robotic manipulation of flexible objects”. Now you are going to get the funds from the European Commission. That is a huge success.
The most important thing is that we can continue what we have already started. We have managed to ensure a smooth transition from Polish to EU funding. This allows us to maintain continuity; we do not have to abandon or withhold any project to focus on the next. What we have achieved with Polish funding can now be used to the advantage of European projects.
The feelings we have had for living beings can be showed to machines
This is also good news for the National Center for Research and Development: the idea of the “Leader” program was to create the leader and that goal has been achieved! Poland’s investment in our efforts has made it possible for me to get the money from the European Commission. What is more, what we presented in the “Leader” project was so highly acclaimed that we were invited to the project involving renowned European centers (the leading one being the University of Bologna). The project is going to take four years, so we have a lot of time. We also have three full-time posts and 750 thousand euros.
What is the project focused on?
Industry is struggling with laying bunches of wires. They are flexible and easily deformable, and robots cannot cope with that. Machines are mostly used to weld, paint, merge, drill or to perform other repetitive operations. But we are not able to position a bunch in such a way so as to ensure that its curvature always stays the same. We want to tackle two-handed manipulation and its purpose – how to make a robot move without cutting or damaging any wires.
The second thing is more intriguing. It is perception, a touch through which we are able to define how a wire may buckle. With that ability, a robot has a chance to figure out how to grab wires of different thickness.
These are just assumptions. The keynote is perception for physical interaction. In other words, we want robots to know what they touch and what they interact with. Most of the systems used nowadays are equipped with 3D sensors. They can determine the shape of the scene, but have no information about whether something gests deformed and resilient or about whether something absorbs energy. The system having no such limitations was designed for the “feet” of ANYmal robot – so as to let it know how it should act, how to change its gait when it runs on an icy or sandy surfaces that absorb energy.
*Krzysztof Walas, PhD – assistant professor in the Institute of Automatic and Information Engineering, Faculty of Electrical Engineering, Poznan University of Technology. He is the author or co-author of 45 publications. He was i.a. the head of organizational committee of the 1st International Workshop on Perception for Mobile Robots Autonomy. Since 2009 he has been a member of IEEE Robotics and Automation Society and national coordinator of euRobotics Week. At his initiative, main events of this year’s European Robotics Week were held at the Poznan University of Technology.
Przeczytaj polską wersję tego tekstu TUTAJ