I study methods for identifying subtle condition differences in small UAVs using multimodal depth, sound, and airflow measurements acquired during controlled indoor hovering.
For each modality, a convolutional autoencoder (CAE) is trained on a reference condition. Reconstruction error is used to quantify condition differences, and the modality-level results are combined for multimodal evaluation.
My work includes experimental setup and data acquisition, sensor placement and measurement-condition design, preprocessing and analysis of measurement data, CAE-based evaluation, and investigation of measurement noise and experimental-condition effects.
