Automated Robotic Picking and Assembly Employing Simultaneous Localization and Calibration

• Main Authors: Shih-Lin Cian, 錢世霖
• Other Authors: Wen-Chung Chang
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/586nuy

碩士 === 國立臺北科技大學 === 電機工程研究所 === 105 === An automated robotic picking and assembly system employing simultaneous localization and calibration is presented in the thesis. The simultaneous localization and calibration approach is used to online determine the transformation between a manipulator and a 3D camera. By employing visual servoing, the manipulator can thus perform automated picking and asssembly with precision. Simultaneous localization and calibration involves simultaneous localization and mapping and the extended Kalman filtering. For the purpose of performing calibration, the transformation between a manipulator and a 3D camera can be determined by recursively estimating and updating the positions of feature points in the world frame with a flying camera. With a 3D camera as the vision sensor, the pose of the objects are computed by using RANSAC (RANdom Sample Consensus), and the pose of the robotic gripper is calculated by image processing. A visual servo controller can thus be designed with appropriately-defined encoded error to drive the manipulator to perform picking and assembly tasks autonomously with precision. The proposed approach has been successfully validated by experiments.