摘要
本文采用YOLOv2深度学习模型对双足机器人拍摄到的第一帧图像进行目标定位检测,接着通过传统的Camshift算法对后续图像中的目标进行跟踪,直到逼近目标后利用机器人关节运动控制程序拾起目标。该方法的优点是:将深度学习算法与传统图像跟踪技术相结合,增强了系统的鲁棒性和实时性。
In this paper,we use YOLOv2 deep learning model to detect a target location through the first frame image captured by biped robot.Then we use traditional Camshift algorithm to track the target in the follow-up image until the target is approached and the robot joint motion control program is used to pick up the target.The advantage of this method is that the combination of deep learning algorithm and traditional image tracking technology enhances the robustness and real-time performance of the system.
作者
刘超强
叶坤
李鹤喜
LIU Chaoqiang;YE Kun;LI Hexi(Faculty of Intelligent Manufacturing,Wuyi University,Jiangmen 529020,China)
出处
《现代信息科技》
2019年第18期137-140,共4页
Modern Information Technology