Automatic Detection of Abnormal Drone Flight State by Measuring Changes in Images Captured by a Drone-Mounted Camera

While in flight, drones can be adversely affected by strong gusts of wind. Currently, drones are equipped with hardware such as accelerometers and speed sensors to mitigate these effects. However, these measures may be inadequate, and intense disturbances can lead to crashes. For instance, a remotely piloted drone blown off course by wind may become uncontrollable, and any delay in correcting its trajectory could result in an accident. Therefore, in this study, we analyze images captured by cameras mounted on drones during autonomous flight to detect unusual movements. This enables the control center to promptly address anomalies, independent of the already equipped internal sensors for anomaly detection, thus enhancing the rapid detection of flight anomalies. Additionally, it acts as a system backup in case these equipped sensors for anomaly detection become nonfunctional. In this article, we assess the drone's speed and direction by analyzing changes in successive images taken by the onboard camera. Based on these assessments, we determine the normalcy of the flight state. For this, we introduce a novel method for detecting abnormal flight conditions, utilizing the observed variations in speed and direction. Extensive experimental testing demonstrates the effectiveness of our proposed method in identifying abnormal flight conditions during drone operations.