UAV Path Planning Based on Biological Excitation Neural Network and Visual Odometer

Abstract

Unmanned aerial vehicle(UAV) have been widely used in military and civil fields due to their compact structure, flexible mobility, low cost and other advantages. With the development of artificial intelligence in recent years, more intelligent and advanced algorithms have appeared, in which machine vision, as an important branch in the field of artificial intelligence, has also been greatly developed. The limitation of space, load, endurance and computing capacity hinders the application of intelligent algorithms on UAV. In the paper a semi-autonomous control platform of the quadrotor UAV was developed and the upper and lower dual control core architecture is implemented. Based on the hardware platform, the improved visual inertia odometer (VIO) and the biological excitation neural network are used to improve the flight performance and the ability of autonomy. To solve the problem of the synchronization for VIO, a cubic spline interpolation function was employed. A biological excitation neural network was extended to solve UAV on-line path planning. It provides an on-board path planning approach for UAV in the 3D world considering the dynamic obstacles. Finally, the feasibility and stability of the designed system were verified by flight experiments.