A new artificial lateral line attitude perception method based on mixed activation function-multilayer perceptron (MAF-MLP)

Attitude perception is significant for state recognition and navigation of underwater applications. However, few studies have combined pressure sensors with gyroscopes for attitude sensing alone, and even fewer have addressed how to obtain the pitch angle, roll angle and travel speed at the same time. To make up for these shortcomings, an artificial lateral line attitude perception method based on mixed activation function-multilayer perceptron (MAF-MLP) model was proposed to recognize various attitudes of underwater vehicles. The artificial lateral line carrier model is created based on the lateral line structure of fish, and its surface pressure distribution characteristics are obtained by simulation analysis. The artificial lateral line (ALL) system and the data collection board equipped with a gyroscope module are used to conduct underwater experiments, and the gyroscope is used to collect the carrier attitude information in real-time to improve model prediction accuracy. Finally, the experimental data agree with the theoretical analysis results, and the MAF-MLP model is used as the data source to predict the carrier attitude. The results show that the MAF-MLP model can accurately predict the pitch angle, roll angle and travel speed, which provides a theoretical basis for realizing the state identification and navigation of the underwater vehicle.