New team of young researchers gets to work

Focus on the interaction between humans and artificial intelligence

2020/10/12

A new group of young female scientists at TU Darmstadt funded by the German Federal Ministry of Education and Research will be conducting research over the next four years into “Interactive AI for Domain Experts and Everyday Users” (IKIDA). The team will be headed by Dr. Dorothea Koert from the Intelligent Autonomous Systems Group in the Department of Computer Science. She was awarded the AI Newcomer Prize in 2019.

Interactive skill learning will enable robots to learn new tasks from everyday users. Dr. Dorothea Koert (left) and Lisa Scherf (right) in the laboratory of the Intelligent Autonomous Systems Group.

The six scientists come from the fields of machine learning, robotics and cognitive sciences. The technical and scientific goal of the project is to develop interactive, probabilistic algorithms for artificial intelligence (AI) that are able to learn and benefit from direct interaction with humans. Findings from this interaction during specific everyday situations should help to make AI solutions more user friendly and sustainable, reduce errors and generate greater trust and acceptance.

The six scientists come from the fields of machine learning, robotics and cognitive sciences. The technical and scientific goal of the project is to develop interactive, probabilistic algorithms for artificial intelligence (AI) that are able to learn and benefit from direct interaction with humans. Findings from this interaction during specific everyday situations should help to make AI solutions more user friendly and sustainable, reduce errors and generate greater trust and acceptance.

The team is also collaborating with the associated industrial partners Energy Robotics (Darmstadt), Franka Emika (Munich) and Porsche Motorsport (Weissach). They will support the scientists in the evaluation of relevant practical applications for the interactive AI methods. These include, for example, the automated classification of data and teaching motor skills to robots. The team will work together closely with the AI initiative “AI:DA” and the Centre for Cognitive Science at TU Darmstadt.

Dr. Dorothea Koert
Dr. Dorothea Koert

“Our main vision is for AI to one day be accessible to everybody”

Interview with Dr. Dorothea Koert

Dr. Koert, you are part of a new group of young female scientists who are about to start conducting research into “interactive artificial intelligence”. What will this research focus on specifically?

We are conducting research in our team into the development of interactive algorithms for artificial intelligence (AI) that will be able to benefit from direct interaction with human users. By enabling this type of direct interaction, the research has the potential to reduce the barriers to the application of AI solutions, increase acceptance for these solutions and improve their sustainability. In the development of these interactive AI algorithms, our group also wants to place a particular focus on creating cognitive models for human-AI interaction and integrating them into the algorithms. The main practical applications of our research will be interactively teaching motor skills to robots, which could be used e.g. assist people in their everyday lives, and interactive AI-assisted data analysis, which could also be an interesting technology for companies.

How did you assemble the team? Which specialist fields are working together in the team?

A strong focus was placed on interdisciplinarity when assembling the team, which is essential in my opinion for finding answers to the research questions that we are investigating. It is fantastic that we have researchers with expertise in the fields of psychology, cognitive sciences, computer sciences, robotics and machine learning in the IKIDA team and I am already looking forward greatly to the active, interdisciplinary cooperation between these scientific fields which will hopefully lead to exciting results.

You are also cooperating with some associated industry partners. Why did you select these companies and what are the common benefits for those involved?

In our opinion, there is a particularly high potential for the use of interactive AI in companies in the future. To ensure that we keep an eye at IKIDA on practical applications from the very beginning, we are looking forward to the ongoing collaboration with the associated industry partners. This collaboration should also help us identify and conduct further research into practical issues that are relevant to companies. The areas being investigated include applications for interactively teaching motor skills to robots and possible application scenarios for automated AI-assisted data analysis. We believe that the planned collaboration with these companies on an ongoing basis can help make the research carried out in the IKIDA more relevant for use in practical applications in the end. Therefore, we are delighted to be collaborating with the associated partners Energy Robotics (Darmstadt), Franka Emika (Munich) and Porsche Motorsport (Weissach).

Finally, let us take a look at your project list and their milestones: What would you like to have achieved with your colleagues in four years?

Our main vision is for AI to one day be accessible to everybody so that the development and adaptation of AI will thus no longer be reserved just for AI experts. As this theme is very complex, however, we will probably only be able to take the first steps in this direction during the four year term of this planned project. By the end of the project, we hope that we will have been able to carry out research into new methods for simplifying human-AI interaction, especially in the areas of interactively teaching motor skills to robots and automated data analysis. In addition, we want to develop and test cognitive models in order to better understand how human-AI interaction in these applications can also by optimally designed for inexperienced users.

The interview was conducted by Jörg Feuck

Statements by the IKIDA team members

Lisa Scherf

I believe that the most relevant aspect of our research in the future will be the development of highly interactive AI systems that will ease the burden on and support people in a variety of different fields – from the care sector through to the automotive industry.

Picture: Katrin Binner

Vildan Salikutluk

My research deals with the interaction between humans and AI, which is not only extremely fascinating but also has great potential for many important applications.

Picture: Katrin Binner

Svenja Stark

I am conducting research in the IKIDA Group because I want to enable humans and robots to learn from one another so that we can understand people better and improve their everyday lives.

Picture: Katrin Binner

Dr. Cigdem Turan

My research area – human-centred AI – is fascinating because we are trying to understand how humans respond, make decisions, interact and socialise; we also use human input to improve the robustness of algorithms and to help people understand machine learning algorithms.

Picture: Katrin Binner

Susanne Trick

I am particularly excited about combining the relative strengths of humans and AI in the best way possible so that they can work together as a successful team.

Picture: Katrin Binner