Research on a visual robot sensor with its sight-line ability.

开发机器人视觉功能作为智能机器人系统的输入设备具有重要的意义。将人眼控制方法，特别是视线控制能力引入机器人视觉系统。我们采用传感器融合的方法对注意区域的图像进行处理，并结合激光扫描装置的立体视觉方法，构建了一个有效执行各种任务的目标识别系统。采用两台CCD相机、激光扫描装置、高速图像处理装置和基于unix的工作站组成了具有视线特征的三维视觉传感系统。该系统通过实时控制输入机器人操作场景，对物体的边缘信息进行处理，获得物体上激光光斑的三维信息。我们还开发了自上而下的识别系统，以实现各种操作目标，旨在由人类操作员识别。该系统具有对智能机器人的多种识别方法进行优化选择和组合的特点。在这个实验中，我们考虑一个具有视觉传感器能力的机器人，通过智能地利用图像处理和激光跟踪方法来抓取一个有抓点的杯子。这种具有传感器融合特征的机器人视觉传感方法为实现更实用的机器人视觉传感系统奠定了基础。

It is important to develop robotic visual function as an input device for an intelligent robot system. We introduce human's eye control method, specially sight-line control ability, to robot's vision system. We contstruct an object recognition system for carrying out various tasks effectively using sensor fusion method for processing an image of the attention area and stereo vision method with a laser scanning device simultaneously.We make 3 dimensional visual sensing system with sight-line features for composing with two CCD cameras, laser scanning device, high-speed image processing device and a uinix-based workstation. This system can input robot operational scene by real time control to process edge information of an abject and obtain 3 dimensional information of a laser spot on the object. We also develope top-down recognition system to realize various operational aims intended to be recognized by a human operator. This system has features of optimaly selecting and combining several recognition methods of an intelligent robot.In this experiment, we consider a robot with visual sensor ability to grasp a cup with a grasping point by intelligently utilizing image processing and laser tracking method. This robot visual sensing method with sensor fusion features provides a foundation for realizing more usefull robot vision sensing system.