Abstract
Robot swarms can provide a new and exciting medium by which humans can express themselves to and through. In turn, a robot swarm needs to express itself to humans, so that humans can understand and be able to read the state of the swarm to interact with it accordingly. Social applications that involve the need for humans to express themselves, such as group-decision making and brainstorming, have not been studied as an application for swarm robotics. In this thesis, we address this gap by designing, building and studying expressive robot swarms to be used in social applications. Furthermore, the area of multi-human, multi-robot interaction is scarcely researched. In our research, many robots are used with many people of varying numbers, hence helping build the foundation of multi-human, multi-robot interaction. These interactions take advantage of swarm characteristics: they are completely decentralised, which makes them robust against single points of failures, they are also flexible to different environments and applications, and they are scalable to large numbers to match the number of humans taking part in these interactions.We first started our research by doing preliminary work with a pre-existing swarm system, Kilobots, to explore human-swarm interaction methods. We built a system, called Robotic Canvas, where a human and a swarm of 220+ stationary Kilobots collaborate to create paintings. We then added movement to the Kilobots and enabled them to detect and represent only salient features in images, thereby decreasing the number of robots required to present an image. With lessons learned from this work about swarm expressivity and human-swarm interaction, we designed and built our own swarm system, MOSAIX, made of up to 100 robot Tiles. Tiles are small touchscreen-on-wheels robots that are easy-to-use by humans and provide an intuitive interaction. We conducted a research study with 46 participants to use the robots in a social task: group decision-making. Results show that the participants were comfortable using the robots and their engagement with the task increased by using the robots. We then took MOSAIX out to the world to use it in 4 more social applications with members of the public. Through our research, and to the extent of our knowledge, we have built the first swarm system that was used in large numbers (up to 63 robots at once) with the public in the real-world in a novel application: social tasks, while maintaining a completely decentralised control.
| Date of Award | 6 Dec 2022 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Paul J O'Dowd (Supervisor) & Sabine Hauert (Supervisor) |