Barometric Altitude Measurement Fault Diagnosis for the Improvement of Quadcopter Altitude Control

Abstract

Due to the influence of airflows surrounding an aerial vehicle, an onboard barometer sensor can provide a significantly inaccurate measurement that results in a considerable altitude flight performance degradation. In this paper, we present a barometric altitude measurement faults diagnosis methodology and use it to improve the altitude control performance of a quadcopter. A feed-forward neural networks structure is proposed to identify the barometric altitude measurement fault model based on the regression neural networks approach. We use real barometer measurement data for the neural networks training and validation process. An altitude compensation algorithm based on the neural network output is implemented in the vehicle's altitude controller. Actual flights were conducted using a quadcopter platform and the experimental results demonstrate the effectiveness of our proposed method.