Multi-Operator Gesture Control of Robotic Swarms Using Wearable Devices

Abstract

The theory and design of effective interfaces for human interaction with multi-robot systems has recently gained significant interest. Robotic swarms are multi-robot systems where local interactions between robots and neighbors within their spatial neighborhood generate emergent collective behaviors. Most prior work has studied interfaces for human interaction with remote swarms, but swarms also have great potential in applications working alongside humans, motivating the need for interfaces for local interaction. Given the collective nature of swarms, human interaction may occur at many levels of abstraction ranging from swarm behavior selection to teleoperation. Wearable gesture control is an intuitive interaction modality that can meet this requirement while keeping operator hands usually unencumbered. In this paper, we present an interaction method using a gesture-based wearable device with a limited number of gestures for robust control of a complex system: a robotic swarm. Experiments conducted with a real robot swarm compare performance in single and two-operator conditions illustrating the effectiveness of the method. Results show human operators using our interaction method are able to successfully complete the task in all trials, illustrating the effectiveness of the method, with better performance in the two-operator condition, indicating separation of function is beneficial for our method. The primary contribution of our work is the development and demonstration of interaction methods that allow robust control of a difficult to understand multi-robot system using only the noisy inputs typical of smartphones and other on-body sensor driven devices.