Finding behavioral patterns of UAV operators using multichannel hidden Markov models


In recent years Unmanned Aerial Vehicles (UAVs) have become a very popular topic in many different research fields and industrial applications. These technologies, and the related industries, are expected to grow dramatically by 2020. Although the systems designed to control UAVs are increasingly autonomous, the role of UAV operators is still a critical aspect that guarantee the mission success, specially when one single operator must supervise multiple UAVs. For this reason, much effort from different areas has been put into the study and analysis of the operator behavior. This work presents a new method to find and model behavioral patterns among UAV operators in a lightweight multi-UAV simulation environment. Our approach is based on MultiChannel (or Multivariate) Hidden Markov Models (MC-HMMs), which allow to gather in the same model parallel data sequences, such as the combination of operator interactions and mission events. The different steps for preprocessing data, creating, selecting and analyzing the model are described, and an experiment with inexperienced operators has been carried out to show how a descriptive model of behaviour can be generated using this modelling technique.

2016 IEEE Symposium Series on Computational Intelligence (SSCI)
Antonio Gonzalez-Pardo
Antonio Gonzalez-Pardo
PhD Computer Science

Lecturer at the Computer Science Department. Main research interests are related to Computational Intelligence and Metaheuristics applied to Social Networks Analysis, and the optimization of graph-based problems.