A methodology for evaluating the training effectiveness of quadrocopter operators based on digital behavioral trace analysis and latent action patterns

The paper proposes and substantiates a new methodology for evaluating the effectiveness of quadcopter operator training Digital Behavioral Trace Analysis, based on end-to-end analysis of the digital trace of small managerial actions. Unlike traditional approaches that rely on objective performance metrics, psychological questionnaires or expensive physiomonitoring, Digital Behavioral Trace Analysis captures high-frequency control parameters: reaction time to a command, frequency of stick micro-corrections, statistics of deviations from the optimal trajectory and additional attributes. To test the methodology, a quadrocopter flight simulator was used, in which 10 operators performed five runs of a standard scenario "takeoff — maneuver — landing". Using automated logging, data were collected for four key attributes, after which the mean values for each participant underwent clustering using the K-means method. The experimental results showed a clear separation of the participants into two groups: "confident" operators with low reaction times and infrequent micro-corrections and uncertain participants with elongated reaction times and high frequency of minor corrections. Identifying latent patterns of behavior without involving subjective assessments or sophisticated equipment provides new opportunities for adaptive simulators and predictive monitoring systems. In particular, DBTA can serve as a basis for automatic adjustment of the complexity of training tasks in real time, timely detection of operator fatigue or overload, and more objective selection of candidates for the position of UAV pilot. In the future, the methodology is planned to be extended to include additional attributes: camera operation, trajectory analysis, interaction with the user interface, and application of supervised models for predicting readiness to perform complex flight scenarios.