Multi-Operator Control of Connectivity-Preserving Robot Swarms

Robot swarms hold the potential to help humans accomplish many challenging problems. Prior research on human-swarm interaction systems has primarily focused on the interaction with a single human operator. This inability to collaborate with multiple humans may limit the swarm’s use in practical scenarios. This thesis considers how multiple humans who are working alongside a robot swarm can effectively use the robots as a shared resource. In an environment that lacks global communication, we consider a scenario where humans freely explore the area to complete spatially distributed tasks. First, we present a framework based on formal languages that allow robots to help a pair of humans complete tasks while preserving the connectivity of the communication network between them. The framework supports automatic code generation to control the robots in a fully distributed manner. Next, we conducted a user study with 52 participants to investigate the usability of the framework under different communication constraints. The participants were each presented with a graphical user interface that provided only a local, first-person view of the environment. Finally, the framework is extended to support an arbitrary number of humans to share the control of a swarm. The framework is first verified in simulation and then through physical robot experiments to demonstrate their application in real life. Results show that the framework creates communication networks that can adapt to the movement of humans, enable humans to exchange robots depending on changing task demands, and are near-optimal with respect to network length and the number of robots required to maintain it. Moreover, they suggest that the ability to exchange robots in the swarm offers humans the flexibility to work independently or collaboratively depending on the current situation. This work can be seen as a step towards designing effective human-swarm teams.