DDCNN: A Promising Tool for Simulation-To-Reality UAV Fault Diagnosis

TL;DR

A novel difference-based deep convolutional neural network (DDCNN) model is presented that uses the difference features extracted by deep convolutional neural networks to reduce the sim-to-real gap and can increase the accuracy from 52.9% to 99.1% in real-world UAV fault detection.

Abstract

Identifying the fault in propellers is important to keep quadrotors operating safely and efficiently. The simulation-to-reality (sim-to-real) UAV fault diagnosis methods provide a cost-effective and safe approach to detecting propeller faults. However, due to the gap between simulation and reality, classifiers trained with simulated data usually underperform in real flights. In this work, a novel difference-based deep convolutional neural network (DDCNN) model is presented to address the above issue. It uses the difference features extracted by deep convolutional neural networks to reduce the sim-to-real gap. Moreover, a new domain adaptation (DA) method is presented to further bring the distribution of the real-flight data closer to that of the simulation data. The experimental results demonstrate that the DDCNN+DA model can increase the accuracy from 52.9% to 99.1% in real-world UAV fault detection.

Figures (8)

• Figure 1: Example of broken propeller.
• Figure 2: Framework of the proposed approach.
• Figure 3: Network struture of DDCNN. (a) The network structure of DCNN_1 and DCNN_2. (b) The network structure of DDCNN used for training. (c) The network structure of DDCNN used for testing.
• Figure 4: Data collection in simulation and real flight. (a) One of the UAV trajectory for data collection in simulation. (b) One of the UAV trajectories for data collection in real flight. (c) Lab scenario for data collection in real flight.
• Figure 5: The real UAV model and broken propellers used in real-world data collection.