Analysis of Anomaly in the Behavior Characteristics of Operators in HSMM-based Multi-UAV Supervisory Control System

Abstract

Solo operation of multi-unmanned Aerial Vehicle (UAV) will be the trend of future, and operators play a significant role in the UAV supervisory control system, but human may be impacted by all kinds of factors, which may decline the performance of UAV system. In this paper, modeling is conducted for the behavioral characteristics of operators in the multi-UAV supervisory control system with improved Hidden semi-Markov Model (HSMM), and the relationship between the predication result and anomaly of HSMM model is analyzed. This method can be applied for monitoring the anomaly in the behaviors of operators, which is of strong and practical significance in the multi-UAV supervisor control system and Human Supervisory Control (HSC) system. The current Unmanned Aerial Vehicle (UAV) system still adopts the manipulation of UAV by several operators. In order to realize the manipulation of multi-UAV by single operator, a lot of studies have been conducted in the automation of multi-UAV control and man-machine work allocation, etc. (1); with the improvement of automation of UAV system, the operation of multi- UAV by single operator will be favourable in the future, and it will also become a significant part in the future network centric warfare (2). There are many differences in the control of UAV by single operator and current UAV control system. In the traditional UAV system, operators mainly control the UAV manually, while in multi- UAV control system, the automatic system takes charges of the majority of operation, and the operators are mainly responsible for the supervision and control of UAV system, as well as the intervention and decision in emergencies beyond the handling of automatic system. Consequently, the future multi-UAV system is a typical Human Supervisory Control (HSC) system (3). In this paper, modeling is conducted for the behaviors of operators in the UAV HSC system with Hidden semi-Markov Model (HSMM) with the experimental data as the drive. When there are anomalies in the behavioral characteristics of operators, the relationship between the prediction result and abnormal behavior of HSMM model is analyzed. With this method, the real-time supervision and analysis of the behavioral characteristic anomaly is implemented in the UAV supervision and control or other HSC scenes.