“Robots have very rough hands”

class=”sc-97fd9fa8-0 jNFKxv”>

Mr. Siegwart, your institute is working on a wide variety of robots. Is there a common denominator?
Roland Siegwart: Our focus is mobile robots that roll or fly. They belong to the new generation of robots: they can actively intervene in the environment, for example, by opening a door. They deal with complex environments and can react to unpredictable situations. Our passion is for drones that not only fly from point A to point B, but also fly into a wall and measure and drill holes there.

Does someone control such an aircraft or independently carry out an order?
The aim is for robots to move autonomously. But I am convinced that in most cases people stay in the loop. Humans are much better than robots when it comes to general decisions – and we will be for a long time to come.

What is the robot better at?
The robot may be better at handling sensitive interactions than humans. For example, an industrial robot performing spot welds.

And where does it defy its limits?
In the long run, tactile interactions are the hardest thing for robotics. Robots have very rough hands.

Does this handicap set boundaries for robots, for example in the medical field?
For example, robots are used in hip operations: They can drill much more precisely. But a robot cannot afford to cut skin and muscle. This requires the tactility of our hands.

That’s why robots and humans are joining forces.
There will always be an interaction. Robots are simply next-generation vehicles.

They emphasize that the robot is a tool and not a replacement for the human, because many are uneasy: It is not yet clear how such systems will change our world of work.
I understand it. Work is something at the center of life. Because robotics is so complex, these changes will take generations. It’s not like industrialization, where textile machinery emerged and large-scale jobs soon disappeared. Robots come in stages. For example, we use them to mow the lawn – no one cares about it.

While you were studying you were still working with punch cards, today your field of research has gone digital. What are the biggest milestones in these decades?
We have to admit, visions in robotics were pretty much the same 30 or 40 years ago. It is the developments in related technologies that enable us to make great strides. For example, the idea of ​​being able to fly the way four- or six-rotor drones do today has been around for a long time. However, you could not implement them because the technology was not available. This requires easily controllable motors, computing power for small systems, and angle sensors. With all this available, we were the first to have such a ship in the air.

Did you fly the first drone?
This was about 20 years ago at my institute at the EPFL in Lausanne. The basic idea was not new, but we were the first to show that it was possible.

It took a generation for the vision to become technologically viable.
There is currently a hoax about deep learning systems. And that basic idea is how to do it with neural networks, a relatively old, primitive copy of how our brains work. But you didn’t have the computing power to do that. It’s different today.

An example of deep learning is chatbots equipped with artificial intelligence, which many are currently experimenting with.
Definitely. I have a skeptical opinion on this. Systems like ChatGPT analyze enormous amounts of data and only then produce good results. A person can create something from relatively little.

What do you want to say?
Showing the child a drawing of a giraffe will give the child an idea about the animal. In the zoo, he will recognize the giraffe even if it looks different from the one in the book. Systems that learn from a large amount of data cannot yet do this.

Recently introduced chatbots are on our computers and smartphones. Will these also be integrated into robots?
Chatbots can be useful in human-robot interaction. A chatbot can convert words into commands for the robot to take action.

Could artificial intelligence become dangerous for us?
As long as they are machines, i.e. dead matter with no reproductive drive, there is no danger unless they are poorly programmed. However, artificial intelligence can outperform us in some tasks. But we’ll be fine with it. As we have seen, a computer can calculate faster than a human.

His team is exploring new and beneficial uses for drones. But in recent months, drones have been used in airstrikes in Ukraine.
This is extremely important to us. All technologies can be used for good and bad. We do not conduct any research in line with military applications. In the field of search and rescue, we only work with the Armasuisse Federal Office of Defense. With drones and other robots able to break into destroyed homes, we have new ways to find victims. Whether in war or in Turkey after earthquakes. There are other places where one should not go.

For example?
People working in mines 2000 meters below the ground have a much shorter life expectancy. A robot should do this job.

Is that what drives you: to do something for the benefit of society?
That’s why I became an engineer. I think you can add a lot with new robot systems. There are many professions where people wear out from working. For example, you will see more robots on construction sites in the future.

They have developed robots that will be used to clean up the old ammunition depot at Mitholz in Bern.
The project started with me, now a colleague is continuing it. Start-up Gravis Robotics uses technology developed at ETH to move excavators autonomously. It should start in a few years in Mitholz.

When I was a kid, in the TV series “Knight Rider”, KITT could think, talk and drive by himself. 40 years later this vision turned out to be very bold. Where is the development of self-driving cars?
My prediction has been stable for 15 years: fully autonomous vehicles will be available in everyday life when at the age of 85 I am no longer allowed to drive and I want to visit grandchildren.

So in about 20 years.
Autonomous driving on well-structured roads has been technically possible for some time and has already been integrated into vehicles. Vehicles are already fully autonomous in restricted areas such as Arizona in the USA, where the weather is always nice. But there is still no autonomous vehicle that can handle a lot of snow or rain.

You talk about technical barriers, but there is also the question of whether the system can make ethically sound decisions.
The same example is repeated over and over: a vehicle must decide whether it is more likely to hit the elderly or the child in the event of an evasive manoeuvre. I say: That’s a question that doesn’t even come up.

From where?
The goal should be for the car to choose the next step in its movement in such a way as to keep the risk of serious or fatal injury as low as possible, whether young or old. Autonomous vehicles will be able to do this much better than humans. When the time comes, I think there will be a debate in society about whether people should still be allowed to drive. Because autonomous vehicles will be safer than humans and cause fewer accidents.

They’re not academics in the ivory tower, they’re trying to turn technologies into products. Explain how to do this at Wyss Zurich, a technology transfer center of ETH, and at the University of Zurich, where you are co-leader.
As researchers at the university, we come up with a lot of new ideas. However, the road to the product is long and arduous. You have to do a lot to establish a company and find investors. Wyss Zurich is an early stage funding option for regenerative medicine, robotics and bionics. The Gebert Rüf Foundation has the same aim.

You are the president of this foundation, which has been supporting start-ups with millions of contributions for 25 years. What is their significance for pioneers in Switzerland?
The foundation has helped us now have a vibrant start-up and entrepreneurial environment in Switzerland, especially at universities. A start-up needs financial support of several 100,000 francs at an early stage. Almost no one is willing to do this because the risk is so high. The Gebert Rüf Foundation covers this.

Can a foundation do this better than the state?
I think so, yes. A foundation can react more dynamically. Start-ups need to be able to react. You are in the middle of a rush. You submit a project, we look at it, you submit it – and the next day you find out if your project will receive funding. Money is available immediately.

Source : Blick