摘要
扑翼飞行器是一种仿生机器人,其翅膀可以像鸟和昆虫的翅膀一样上下扑动。本文提出了一种基于视觉的扑翼飞行器避障方法。首先,使用稠密光流算法计算机载图像传输摄像机拍摄的第一视角视频帧的光流场。基于所获取光流信息作为输入量,设计模糊避障控制器来给出扑翼飞行器的导航指令。实验结果表明,本文所提出的避障方法能够准确识别障碍物,实现扑翼飞行器的避障。
The flapping-wing air vehicle(FWAV)is a kind of bio-inspired robot whose wings can flap up and down like bird and insect wings.A vision-based obstacle avoidance method for FWAVs is proposed in this paper.First,the Farneback algorithm is used to calculate the optical flow field of the first-view video frames taken by the on-board image transmission camera.Based on the optical flow information,a fuzzy obstacle avoidance controller is then designed to generate the FWAV steering commands.Experimental results show that the proposed obstacle avoidance method can accurately identify obstacles and achieve obstacle avoidance for FWAVs.
作者
付强
王进
宫乐
王靖元
贺威
FU Qiang;WANG Jin;GONG Le;WANG Jingyuan;HE Wei(School of Automation and Electrical Engineering,University of Science and Technology Beijing,Beijing 100083,P.R.China;Institute of Artificial Intelligence,University of Science and Technology Beijing,Beijing 100083,P.R.China)
基金
This work was supported in part by the National Natural Science Foundation of China(Nos.61803025,62073031)
the Interdisciplinary Research Project for Young Teachers of USTB(Fundamental Research Funds for the Central Universities)(No.FRF-IDRY-19010)
the Beijing Top Discipline for Artificial Intelligent Science and Engineering,University of Science and Technology Beijing.
关键词
稠密光流
单目视觉
避障
扑翼飞行器
模糊控制
dense optical flow
monocular vision
obstacle avoidance
flapping-wing air vehicle
fuzzy control
