Forensic Analysis of Drone Malfunction Based on Location Data

A drone malfunction occurs when the drone cannot operate normally. This situation may bother the pilot when carrying out missions. Several of the drone malfunctions are caused by the hardware and sensors. To examine the reasons behind drone malfunctions, various strategies are available, with deep learning being one of the notable methods. This research presents a novel deep learning approach, the GPS-Transformer model, for the precise detection of drone malfunctions through the analysis of GPS location data. By leveraging the Transformer architecture, this model effectively captures spatiotemporal patterns within drone trajectory data, surpassing conventional machine learning methods and other deep learning models in accuracy, precision, and recall. The model's interpretability, achieved through attention mechanism visualization, enhances its utility in safety-critical scenarios, empowering operators to comprehend the rationale behind malfunction detection decisions. This advancement holds the promise of significantly enhancing the safety and reliability of drone operations across a wide range of industries, from agriculture to surveillance and beyond. The research shows the Transformer has outperformed accuracy to classify drone malfunctions based on GPS data, with an accuracy value of 96.3% and an F1 score of 95.4%.