Real-Time Detection and Diagnosis of Flight Anomalies in Small Unmanned Aerial Systems

Safety critical systems, such as small Unmanned Aerial Systems (sUAS) must be monitored closely to identify, diagnose, and potentially mitigate flight problems as they arise. In the pre-flight stage, safe planning and sequencing of mission tasks is a crucial job considering the roles of individual sUAS along with their associated multi-vehicle, cross-role dependencies. Also during the flight, the multivariate time series data typically generated by sUAS flight controllers can be complex to understand and analyse. While formal product documentation often provides example data plots with diagnostic suggestions, the sheer diversity of attributes, critical thresholds, and data interactions can be overwhelming to non-experts who subsequently seek help from discussion forums to interpret their data logs. Solutions based on deep learning or heuristics can be used to detect anomalies in different time-series data attributes. However, understanding and mitigating the root cause of flight problems based upon the combination of multiple detected data anomalies requires significant domain expertise. This dissertation addresses two primary problems: (1) designing effective and efficient anomaly detectors to detect anomalies in the sUAS flight data for both real-time and post-mortem analysis, and (2) based on the combination of multiple detected data anomalies, diagnosing the root cause of the problematic behaviour. In addition it also addresses a secondary problem that occurs when faulty sequences of tasks are assigned to a fleet of sUAS.