摘要
提出一种基于图像的运动序列生成技术 ,此项技术不需要已知目标物体及所在背景的 CAD模型 ,也不需要对摄像机进行标定。在样本图集的基础上 ,通过 Karhunen- L o-eve(K- L)变换得到图像的特征表示。建立了机器人关节角与图像特征之间的非线性关系。在机器人关节空间进行轨迹规划 ,最后生成关于机器人虚拟运动的图像序列。以 UP6机器人为目标物体 ,进行虚拟运动实验 ,得到 UP6机器人的虚拟运动 。
A technique is developed to synthesize movement sequences from images without CAD models of the objects or the scene and without camera calibration. The Karhunen Loeve (K L) transform is utilized to acquire image features from sample images. This paper establishes the nonlinear relationship between robot joints and image features. A joint space trajectory is first generated and used to synthesize the robot movement sequence. The virtual movements of UP6 manipulator are obtained based on the experiments. The results show that the technique is convenient and accurate.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2003年第4期466-469,共4页
Journal of Tsinghua University(Science and Technology)
基金
国家"八六三"高技术项目 ( 863 -2 -4 )
