SwarmMonitorAI: Design of AI-Assisted User Interface for Drone Swarm Monitoring
Term
4. Term
Publication year
2024
Submitted on
2024-06-07
Pages
16
Abstract
With the increasing popularity of drones and the advancement of AI, the field of Human-AI Interaction is at risk of violating traditional usability guidelines for user interface design. In continuation of research done by the HERD Project, this Master’s Thesis explores how a design for an AI-assisted user interface can help reduce the user’s mental workload while monitoring live data from a drone swarm. To address this, we took a design approach and created the SwarmMonitorAI to which we designed three prototypes representing three levels of AI assistance: a baseline design, an AI-assisted design, and a design with incorporated autonomy. Subsequently, we conducted a comparative experiment with 20 participants, in which we measured the subjective mental workload with Raw TLX sheets, performance through success rate and error rate, and qualitative insights into the experience through post-task and debriefing questions. As the AI-assisted design had the best results with a 10.50% error rate, 87.25% success rate, and the lowest perceived workload, we redesigned the SwarmMonitorAI accordingly while incorporating elements from the other prototypes that were well-received by the participants. The redesign included improvements to UI elements, user-requested functionality, and a focus on passive assistance features.
Keywords
Documents