旋翼飞行机器人多机视觉跟踪定位系统及方法  

Visual tracking and positioning system and method of multiple rotorcrafts

作　　者：王耀南[1,2] 罗琼华 陈彦杰 谭建豪 刘芯[1,2] Wang Yaonan;Luo Qionghua;ChenYanjie;Tan Jianhao;Liu Xin(College of Electrical and Information Engineering, Hunan University, Changsha 410082, China;National Engineering Laboratory for Robot Vision Perception and Control Technology, Changsha 410082, China;College of Mechanical Engineering and Automation, Fuzhou University, Fuzhou 350108, China)

机构地区：[1]湖南大学电气与信息工程学院,长沙410082 [2]机器人视觉感知与控制技术国家工程实验室,长沙410082 [3]福州大学机械工程及自动化学院,福州350108

出　　处：《仪器仪表学报》2018年第2期1-10,共10页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(61573134,61433016,61471167);国家科技支撑计划(2015BAF13B00)项目资助

摘　　要：针对旋翼飞行机器人在多机跟踪定位时受遮挡、光照变化等因素的影响,易出现跟踪目标丢失的问题,提出一种改进的分段多目标跟踪定位方法,首先,为克服单一颜色特征不能准确描述目标的不足,设计了按一定的权值系数融合目标的颜色特征和边缘特征来建立似然函数作为目标模板;考虑到粒子滤波算法能解决目标发生遮挡的跟踪问题,然后,利用目标模板与候选目标模板的相似度来实现连续自适应Mean-Shift算法和粒子滤波算法的切换,保障了算法的鲁棒性及实时性。可应用于旋翼飞行机器人完成自主的直线飞行与绕圆飞行任务。During multi-target tracking and positioning,the rotorcraft is influenced by the occlusion,illumination changing,etc. These factors will result in losing the tracked target. To solve this problem,a segmentation approach of multi-target tracking and positioning is proposed. Firstly,the shortcoming of single color feature inability to accurately describe a target is considered. The color and edge features of the target are used to establish likelihood function for tracking target according to a certain weight coefficient. Meanwhile,the particle filter algorithm which can solve the tracking problem of target occluding is also utilized. Then,camshift and particle filter algorithm are switched by using the similarity degree between target template and candidate target template. In this way,the robustness and real-time ability of the algorithm are guaranteed. The proposed approach can be applied to control the rectilinear and round flight of the rotorcraft.

关 键 词：旋翼飞行机器人 连续自适应Mean-Shift 粒子滤波 多特征融合 视觉定位 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置] TH741[自动化与计算机技术—控制科学与工程]