Obstacle Avoidance of UAV Based on Neural Networks and Interfered Fluid Dynamical System

Obstacle avoidance is the prerequisite guarantee for the unmanned aerial vehicle (UAV) to fly safely in the three-dimensional dynamic complex environment. In this paper, a three-dimensional real-time obstacle avoidance method is proposed by combining neural network and the Interfered Fluid Dynamical System (IFDS) for the first time. First, in order to solve the problem of insufficient samples, sample data are generated based on the sparrow search algorithm (SSA) and receding horizon control (RHC). Second, training neural network offline, the relative position between UAV, destination and obstacle from sample data as input of neural network, and the IFDS parameters are used as the feature extraction of the output terminal of the neural network. Third, the trained neural network is used to adjust the coefficients of the IFDS according to environment in real time. Finally, the simulations demonstrate effectiveness of the proposed method.