A Bionic Intelligent Obstacle Avoidance Method for UAV Based on Optical Flow

Abstract

The invention discloses an optical flow-based intelligent bionic obstacle avoidance method for a drone, which realizes the obstacle avoidance flight of the drone through the combination of software and hardware, and belongs to the fields of artificial intelligence and computer vision. The specific method is to capture the image sequence through the monocular camera on the drone, and use the LK (Lucas Kanade) method to calculate the optical flow field according to the two frames of images before and after, and filter the noise optical flow; divide the image into edge, left middle, The three parts in the middle right, based on the SVM classifier, determine whether the left and right sides are large obstacles; calculate the TTC sum of each column, calibrate and interpolate the free and obstacle intervals, and obtain the largest free connected interval; combine the left and right optical flow balance strategies to obtain the translation amount and the amount of rotation to control the UAV obstacle avoidance flight. The invention can realize the real-time obstacle-avoiding flight of the unmanned aerial vehicle in the unknown messy environment.

Description

technical field [0001] The invention relates to a monocular vision-based optical flow-based intelligent bionic obstacle avoidance method for unmanned aerial vehicles, and in particular belongs to the fields of artificial intelligence and computer vision. Background technique [0002] A UAV is a power-driven aircraft that can autonomously control flight according to a predefined trajectory or be controlled in real time by remote control. It is more and more widely used in military and urban applications, but in practical applications, most of them rely on real-time remote control by professional operators. The cost of use remains high, the response speed is slow, and it is limited by the transmission quality of communication channels. Therefore, there is an urgent need for UAVs to have the ability to autonomously avoid obstacles to identify and avoid obstacles, so that UAVs can reach the target position in time and accurately, and complete the predetermined function. The eff...

AI Technical Summary

Problems solved by technology

At present, the behavior patterns and neural structures of flying insects have been clearly understood, and some attempts have been made on simulation platforms and ground mobile robots. The application of drones is still in its infancy, and in-depth research is urgently needed. There are still great challenges and research prospects, and it has great scientific significance

Images