机器人相机空间智能编程方法研究*

针对传统的示教编程方式存在操作复杂，效率低，危险性高等不足，严重限制了工业机器人的推广应用。基于自然的人机交互示教方式，提出了一种基于计算机视觉的相机空间工业机器人智能虚拟编程方法，本方法不需要实际操作示教盒和机器人本体，仅采用辅助示教工具在视觉相机空间示教就实现了工业机器人的虚拟编程。主要研究了实现该方案的关键技术即基于相机空间映射模型的视觉定位技术以及基于K-means聚类算法实现的相机空间映射关系自学习技术。最后，基于自主开发的机器人平台，开展基于相机空间的虚拟智能编程实验，验证了本文提出的相机空间工业机器人智能编程方法的可行性及正确性。

Intelligent programming for industrial robot based on camera space

The existed traditional online teaching programs have various problems such as complicated operation, low efficiency, high-risk, long time training which are limited for the application of industrial robot. Therefore an intelligent programming approach which is using an auxiliary teaching tool in the visual space of the camera based on natural human robot interaction is proposed without operating the teach pendant and robot itself. The proposed method relies on the implementation of the visual positioning using camera-based space mapping model and the camera space mapping is clustered by intelligent self-learning using camera space model and the K-means clustering algorithm. Also experiments have been conducted on our developed robotic platform and the results verified the feasibility of the proposed method.