| Summary: | 碩士 === 國立中山大學 === 機械與機電工程學系研究所 === 106 === Humans can achieve many tasks through the mechanism of hand-eye coordination. This study will incorporate a robotic manipulator with a video camera and provide the robotic manipulator with human-like hand-eye coordination capability to accomplish the mission of pouring water. Image processing techniques will be applied to the image information acquired by the camera. This research aims to lay the foundation for motion control of future intelligent robots.
When humans perform the pouring operation, they will first observe the distance between the pouring cup and the target container. Then they will move the pouring cup to an appropriate position and start rotating it. In order to allow the water to be poured into the target container smoothly, the amount of flow needs to be controlled by adjusting the inclination angle of the pouring cup. In this study, a Kinect V2 camera is employed to capture the image. First of all, the robotic manipulator is commanded to a target location by solving the actual distance between the target container and the robotic manipulator using the approach of image processing. Then an algorithm based on physical properties of flowing water is developed to estimate its width and a flow controller is applied to control the amount of flowing water by regulating the inclination angle of the pouring cup. As a result, under the condition of no spilling, the water in the pouring cup can be successfully transferred to the target container with a stable flow rate. Experimental results indicate the proposed intelligent control scheme is able to accomplish the task of pouring water with up to 95% of water. Although the performance is a little bit less than 98.5% done by humans, human-like dexterous skills of hand-eye coordination have already been demonstrated.
|
|---|
