Towards a Multi-Function Swarm That Adapts to User Preferences

Abstract

This paper describes the current state of our work towards a multi-function swarm of UAVs that adapts to user preferences. To achieve this, we employ software simulation and an evolutionary search method for offline generation of a repertoire of controller parameters, covering a wide range of swarm behaviors. By allowing the user to select the appropriate controller parametrization online, the swarm system can adapt to human preferences at run time. To validate our approach, we have performed real-world validation of a few select controllers, on a small fleet of modified commercial off-the-shelf UAVs. The results so far are encouraging, and we plan to extend the range of swarm behaviors and increase the swarm size in the future.