Quantifying collaborative strategies and identifying performance breakdowns of UAV C2 teams using multidimensional cross-recurrence quantification analysis

Abstract

We analyzed eye tracking data of pairs of participants using Multidimensional Cross-Recurrence Quantification Analysis (MdCRQA) metrics, a set of nine metrics that quantify the dynamic relationship between two time series over time. Eye tracking data was collected from twenty-six pairs of participants working together on Unmanned Aerial Vehicle (UAV) command-and-control (C2) tasks where workload increases from low to high. The findings through MdCRQA showed that when workload increased, pairs tend to exhibit more rapid transitions and frequent individual shifts in focus between different Areas of Interest (AOIs). Significant reductions were observed in several MdCRQA metrics, including Average Diagonal Line Length (L), Maximum Diagonal Line Length (MaxL), and Diagonal Line Entropy (EntrL), indicating a shift towards more efficient division of labor and more predictable shared attention patterns. Correlation analyses between the MdCRQA metrics and performance measures revealed that higher values of the aforementioned metrics in high workload conditions were associated with improved response times, suggesting that effective visual coordination is critical for task efficiency under increased workload. The findings here suggest that MdCRQA can provide a subset of metrics that: (a) are sensitive to workload changes, (b) can be indicators of performance breakdowns, and (c) can quantify how teammates collaborate and adapt to workload increases in UAV C2 operations.