AN ADDED VALUE FOR THE PEOPLE
In an interview, professor Oliver Korn takes a look into the future of robots.
Coming into Oliver Korn’s office, first thing, the visitor meets Pepper. The little guy certainly looks nice and friendly, but the interaction with the professor is still a bit of a struggle. But Korn is working on improving that. In an interview with the Mittelbadische Presse, the expert in robotics talks about the possible use of robots in patient care, the hand as a miracle of nature and when the machines will eventually take over.
About Oliver Korn:
Oliver Korn received his doctorate in computer science at the SimTech Excellence Cluster of the University of Stuttgart. His work on assisting systems was awarded the Gips-Schüle Research Award. In 2015 he started as professor for Human Computer Interaction at Offenburg University. In 2017 he founded the ACI (Affective & Cognitive Institute). He also is co-founder of the software development company Korion GmbH, a Fraunhofer spin-off.
Professor Korn, what exactly is a robot?
The term actually comes from a novel and is a derived from of the word for work. That is why, especially in Germany, most people first think of an industrial robot, i.e. a working robot, when talking about the topic.
What we, including my colleague Mr. Hochberg with his football-playing SWEATY, are doing with social robots is something quite different. We are focusing on “social robots” that have to interact and communicate with people. This is what, by definition, distinguishes them from industrial robots, which are actually only machines with sensors.
What are robots currently capable of?
Physically, they are capable of a wide range of applications. We could build a robot able to lift a house. Some of the tools we have, for example the tunnel drills with their sensors, are like massive robots. What defines a robot is a certain degree of autonomy. But if it is artificial intelligence you’re talking about, then robots are still severely limited.
Why is that?
Robots only work well if the application is relatively specific. For example, “drive into a burning house to extinguish the fire”. Now it could use thermal imaging cameras to identify the source of the fire and act accordingly. But imagine a situation where a robot enters a room with an ill person. There are a so many variables that play into the equation: maybe there are guests, maybe the furniture was moved, maybe there are pants over the chair that had not been there before. Maybe the patient is in a bad mood, maybe he’s in a good mood. This is all information that humans process immediately using social skills that have been developed over decades. A robot, on the other hand, must be tediously trained to be able to process this information. So, the more complex the situation, the more difficulties the robot will have to master it.
Is it the hardware or the software that limits the abilities?
What makes a robot so expensive is the hardware. Take the hand for example. As of today, it is already possible to recreate it mechanically to the point where it can grab objects. But before we have robot hands that can actually compete with the human hand, I suppose another 100 or 200 years will have passed. It is almost impossible to reproduce this fantastic combination of power and sensitivity. Mentally however, I wouldn’t be surprised to find a robot that can entertain or interact with you at almost human level much earlier. However, if you shook its hand you’d recognize the difference immediately.
Why is it so challenging to build a hand?
The hand is a highly integrated organ. You have several components in one finger, the bones are all connected to each other. Nevertheless because of the strong muscles you are still able to clench it really tight. You can gently stroke a baby’s head with your hand or cut down a tree with an axe. This is a huge spectrum. Such flexibility simply does not exist in robots. A robot is built either for chopping wood or for stroking a baby. But never for both.
What are your most exciting projects at the moment?
Our research focuses on technology supporting or assisting people. So far, robots often are not the ideal solution. We are working on assisting systems for people with cognitive disabilities, for example. These systems use sensors similar to those used in robots as well as similar AI technology to recognize and evaluate actions, for instance. However, these systems are permanently installed above a workbench. There is motion detection and a projection that tells people: here you have to pick this up and this is what you have to do next. Together with Fraunhofer , we are currently developing a kind of interactive mouse that guides the worker’s hand to the right spot.
Here it is important to find out what robots should be capable of by now – for example how well they are accepted in the health and care sector. Discussing this topic, I recently visited the Careum Forum in Zurich. Interestingly, the opinions were quite mixed: about half of the care experts are against using a robot to change diapers, for example.
Would that even be technically possible?
I think it is technically possible, even if it would require a lot of sensor technology. After all, these are highly sensitive parts of the body. However, such a robot would be rather expensive, and it would probably be a long time before it would be more cost efficient than a nurse. On the other hand, finding enough young people for the care profession is a highly-discussed problem. And activities like diaper changing and heavy lifting might scare off some. The use of robots could contribute to making this profession more attractive. In general, I have observed that the acceptance of such form of personal support is greater among younger caregivers than among older ones. The younger ones argue that a robot would give them the freedom to interact more closely in other ways with the patient, e.g. by small talk.
Do you see any limits to the use of such robots?
Difficult. Dementia and Alzheimer’s patients for example show certain automated behavior and say the same things over and over again. A robot that adopts such conversation patterns could be of great value for these people. From an ethical point of view of course, you might call that a fraud. And real small talk should really not be done by robots. But these people are cognitively incapable of that. It remains an ethical dilemma.
Why aren’t such robots used more often. Is this due to high costs or to the robots’ skills?
Robots are currently still very expensive to buy and maintain. There are two people in our team who constantly make sure that all systems are working. A maintenance contract can cost several thousand euros per year. Thus, robots are still not very efficient for the end consumer market. The only exception are vacuum cleaning robots. However, their cognitive abilities are still at a very low level.
In what direction is the development heading?
I think in twenty or thirty years, when robots in the domestic environment developed more skills, you will probably be very pleased that your bed has been made or other simple tasks have already been carried out for you. The garbage may not be emptied because stairs are a challenge, but maybe it is tied up and all ready. A household robot could do the same jobs a servant carried out a hundred years ago.
For who would such a robot be of special interest?
At first mainly for wealth persons. I know people who would spend 50,000 or 100,000 Euros for a robot that would help them around the house – as much as for a car. I am convinced that this market is very interesting. Players like Sony and Apple will soon enter the market as well.
Is the development in this area taking place gradually or will it go step by steps?
There is an IEEE webpage that lists all currently available robots. New ones are constantly being added. That’s why I would say it is developing gradually.
Technology develops in cycles. Where in this cycle are we right now?
It all starts with a technological impulse, then there is a hype, followed by a peak of inflated expectations. Then comes the valley of diappointment and slowly it heads up to the plateau of productivity. I’d say we’re still between the exaggerated expectations and the valley of tears. And in order to reach the productivity plateau, people will have to be prepared to spend 5000 euros on a robot that performs certain tasks, such as solely taking out the garbage. Unfortunately, we are not yet at that level. Maybe in 10 or 15 years we will be ready for the first people to buy such robots as a kind of toy and then they may find their way to the mass market.
How do you see Baden-Württemberg’s companies positioned in this dynamic market?
There is Kärcher in the consumer sector. In the industrial sector, Festo, Bosch and Mahle are big players – but there are also many small and medium-sized companies. So, there is already something going on here. However, on the level of social robots that interact with people, there are not that many players that come to my mind. At the moment these mostly are smaller companies that build on existing platforms like Pepper. And a lot is still happening in the labs only.
Another important aspect of your work is gamification. What is it about?
Gamification is when you combine regular activities with playful elements. For example, there are apps that reward you with points if you clean your house. Or do you remember the TV show “Eins, zwei oder drei” with Michael Schanze? The children received a ball for every correct answer: this is the gamification of education.
Are there any other examples that come to mind?
Sure. All those fitness watches that tell you whether you have reached the 10,000 steps for today or not. That is the gamification of everyday life. Climbing stairs is not just climbing stairs but also a competition. Through my fitness watch I can stay in touch with my buddies, who will write me a message like: ‘Hey what’s going on couch potato, you’re not moving at all.’ This is a playful enhancement that can be used in various ways. Another example is a European project where we develop an interactive vest for deaf-blind people which helps them to navigate via vibration. Learning how to use this system is supported by playful elements.
This question has to be asked: when do the machines take over?
Depending on how you define power. See, almost everyone uses cars to get from A to B. So basically, cars seem to be in control when it comes to getting to places. People have always used tools. But they remained tools we were controlling. When it comes to the question when the tools will take over the decision making, I think that will not happen for a very long time. The systems are just not ready yet. That’s also more of an artificial intelligence problem anyways.
Would it scare you if an artificial intelligence suddenly started to make decisions for you?
When I look at current developments in society and politics, in Germany and on an international level, an AI would have to be pretty undeveloped to get us to the same situation that is currently going on in politics. So, it might even be desirable that political decisions are no longer made by humans, but by an artificial intelligence. The acceptance would of course be a problem. For me, a good AI wouldn’t be the such a nightmare. Just take a look at the situation in Syria: The horrible things we are seeing right now are the result of many wrong human decisions.
Isn’t it crucial whether we are still able to limit an artificial intelligence?
At the moment there is a lot of research concerning “Explainable Artificial Intelligence”. Decisions of an AI must be explainable. This is challenging, because neural networks operate completely different than our brain. We have to make sure an artificial intelligence is not just optimized for efficiency: otherwise it might come up with the idea to eliminate all humans in order to solve climate change for good. We are indeed the main cause of carbon dioxide, and without us the planet would finally get better. Before artificial intelligence is allowed to make political or societal decisions, we have to ensure that they are in the best interests of humanity.