| Summary: | 碩士 === 南臺科技大學 === 電機工程系 === 107 === This thesis develops an eye-to-hand configuration to calibrate the stereo cameras. Firstly, the left and right cameras are used to capture the image of the object. The regions with convolutional neural network (R-CNN) and image processing algorithms are used to extract the specific features of the object. Secondly, the triangulation method is used to calculate the pose of the object in the camera coordinate system. The transformation between camera system and robot arm coordinates are converted by using an adaptive network-based fuzzy inference system (ANFIS) to reduce computational complexity and maintain system accuracy and then calibrated to estimate the pose of the object in the robotic arm coordinates. Finally, the end effector of the robotic arm is driven to pick up the target object and place it at the desired position in the work area.
Several experiments have been carried out in this thesis, including stereo camera calibration, 3D object pose estimation, eye-by-hand configuration calibration, R-CNN object recognition, and object picking tasks using a 6-axis DOF robot arm with the end effector to demonstrate the proposed method and its effectiveness. The results show that the developed stereo camera calibration obtained the external and internal parameters required during the triangulation process. 3D pose estimation and R-CNN have successfully detected object position and pose in the camera coordinate system and marked the object. In addition, eye-by-hand calibration converts the pose of the object in the camera into a robotic arm coordinate system. Finally, the object picking and placement task was performed correctly using the 6DOF robot arm.
|
|---|
