Explainable Swarm Programming through Imitation Learning

Focus topic:	Artificial Intelligence
Type of funding:	Individual funding programmes
Programme:	CZS Nexus
Funded institution:	Universität Konstanz

Dr. Jonas Kuckling researches the development of software for robot swarms, e.g. using imitation learning. He studied computer science at the University of Paderborn and completed his doctorate at the Université Libre in Brussels. He has been a postdoctoral researcher at the University of Konstanz since April 2023.

Goals

Robots of the future will have to work together in situations and groups that could not have been foreseen during their development. They must therefore be able to learn how to interact with their environment and with other robots on the fly.

In his research project, Dr. Kuckling is investigating how robot swarms can learn collective behaviors. In doing so, they imitate demonstrations or recordings of desired collective behaviors. This allows the robots to develop models of the behavior of other robot groups they encounter and thus effectively learn how to interact with them. In addition, this approach can be used to train robot swarms to mimic natural collectives, such as locust swarms or huge groups of fish. By mimicking such systems through robot swarms, they can be further studied and additional insights into the fascinating behaviors of animal swarms can be gained.