DusseldorfWhether in rescue operations or in the exploration of difficult terrain – learning robots will be able to support rescue workers even more in dangerous environments in the future. This is clear from the report from the “Hostile Environments” working group on the Learning Systems platform, which will be published this Thursday. The platform was founded in 2017 by the Federal Ministry of Research.
Learning systems could provide even greater support, protection and speeding up of rescue operations in the event of terrorist attacks, chemical accidents, major fires and natural disasters, write the authors of the report.
But how far are we from the ultimate “rescue robot”, which replaces human rescue workers with life-threatening dangerous situations and deprives them of life-critical judgments? Hauke Speth, fire director of the city of Dortmund and member of the working group “Lebensfeindliche Environments” of the platform “Learning Systems”, sees such future scenarios skeptical. The development of rescue robots is still in its infancy. But he is convinced of the supportive use of AI-controlled robots in dangerous situations.
“If you detach yourself a bit from the sci-fi vision of a robot that can do more or less everything yourself and run around uncontrollably, then it's clear that robots can close any gaps that still exist today and shorten deployment times.” says Speth.
He believes that robots and humans will save lives together in the future. However, the robot will always be dependent on humans in rescue missions. Because such missions are also about existential questions.
If, for example, a firefighter enters a burning building, he must weigh in a very short time who he will rescue first and if he has to leave someone behind in case of doubt. For such “trolley dilemmas”, when using artificially intelligent robots, a formal framework must be found that incorporates technical, legal and ethical levels, the authors of the report write “Learning Systems in Environ- mental Environments”.
Speth believes the AI of rescue robots is reaching its limits at this point. “There are certain decisions that can never be made by a machine,” he says. “Every single situation would have to be mapped with all kinds of influence. Therefore, I do not believe that the question of who should be saved first can be mapped by an algorithm. “
Answering the question of responsibility is also difficult: Who is at fault if rescue robots make mistakes and people die because of this? It will always be the person who makes the final decision in the future, Speth believes. For this it is necessary that he could intervene at any time in self-learning systems and robots.
Instead of the completely autonomous rescue robots, according to Speth, the development will therefore result in a cooperative system between man and machine. Autonomous flying drones are already creating situational images of job sites much faster than humans. In the future, robotic learning robots will also be used even more in areas that are difficult to access, for example after accidents involving radioactive radiation or toxins. After earthquakes, they will increasingly search for survivors in unstable buildings or debris without a rescue worker having to risk their lives.
However, in the future, it will continue to be people who will monitor these robot missions to intervene quickly in the event of “trolley dilemmas”.
More: Digitalization is increasing the demand for moral algorithms. Businesses need to prevent their systems from interfering with the sound or harming people.
Whether it's Artificial Intelligence or Blockchain, the speed at which new buzzwords from the digital world pounce on us is enormous. But what has substance, what is hype? In our new multimedia section “Digital Revolution”, Handelsblatt editors highlight how digitalisation is changing our businesses, our society and our everyday lives. Each week, the editors take on a focal topic. The various aspects are presented in multimedia form in several contributions. It uses interactive graphics, videos or picture galleries.