This research paper discusses a new technique for diagnosing faults in unmanned aerial vehicles (UAVs), particularly in windy environments. The proposed method, named the uncertainty-based fault classifier (UFC), uses an ensemble of difference-based deep convolutional neural networks (EDDCNN) to reduce model variance and bias. The UFC is designed to filter out uncertain predictions, thus enhancing the accuracy of fault diagnosis in real-world scenarios. The study shows that the UFC can achieve 100% fault-diagnosis accuracy with a data usage rate of 33.6% in windy outdoor scenarios.
Publication date: 22 Sep 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2309.11897
