Innovators under 35 prize for AI project
Technology Review honors the ten winners of the Innovators under 35 competition and presents their ideas for the future. A four-member jury awarded Daniel Leidner for his work in the field of artificial intelligence in robotics. He works at the German Aerospace Center (DLR) with the humanoid robot Rollin’ Justin on the research projects METERON and SMiLE.
Daniel Leidner, 32 and a scientist at the DLR Institute of Robotics and Mechatronics, accepted the award in Berlin on September 6, 2019. For six years, Technology Review Germany has organized the Innovators under 35 competition to find the best ideas of young people.
The interview (excerpt) with our winner:
The research projects you are mainly involved in are called METERON and SMiLE – what are they about?
The two projects combine the idea of controlling robots via a tablet or a smartphone. What was originally intended for astronauts like Alexander Gerst on board the International Space Station with METERON has proven to be extremely intuitive. Therefore, we will now investigate how the findings gained in service robotics can be used for people in life situations with limitations (SMiLE).
Daniel, you taught robots to act autonomously. How did you get the idea?
Autonomy is the chief prerequisite for future robots to help us with dangerous, repetitive and difficult tasks. When I wrote my Master’s thesis at DLR, Justin was only able to do simple tasks following predefined sequences. It was clear that Justin would have to become more independent to be helpful in space and in everyday life.
How do you teach Rollin’ Justin the key tasks? Where’s the AI? And what is the robot able to do?
Justin’s artificial intelligence is based on a knowledge base that manages interaction patterns for known objects. Semantic planning allows the robot to act autonomously with this information; the crew only has to transmit abstract commands using a tablet app. Last August, the German astronaut Alexander Gerst used this method to solve the most complex tasks to date, which had previously been carried out in cooperation with a remote-controlled robot. Not only was Gerst able to clean a solar panel and mount an antenna, his robotic colleague even allowed him to exchange a burnt-out service module – without any training for such emergencies.
This is just an excerpt from the interview. The complete interview will be available on the central DLR website on September, 9, 2019.