Guidance of 5-axis hybrid delta robot with RGB-D camera using visual servo approach

Integrating image processing and artificial intelligence algorithms into robotic systems has significantly transformed modern manufacturing processes. Modern robotic systems are capable of perceiving their environments and making decisions thanks to deep learning models and computer vision techniques, unlike traditional robotic systems. In traditional robotic systems, the position or orientation of workpieces must remain fixed, and any change in the position or orientation of either the robot or the workpieces requires reprogramming the robot. Such constraints create significant disadvantages, especially in terms of time efficiency and labor productivity, and lead to reduced flexibility in production processes. Therefore, the adoption of innovative and intelligent robotic techniques has become increasingly essential. In this thesis, a five-axis hybrid delta robot kinematic model is developed to facilitate autonomous operation through a position-based visual servoing method. The robotic system is designed to perform effectively despite changes in the position and orientation of the robot workpieces, thereby eliminating the need for manual reprogramming. Delta robots are commonly used in industrial applications due to their high speed and high accuracy advantages. In this context, a new kinematic model of a hybrid delta robot has been developed with a focus on low cost and high accuracy, and its functionality has also been tested under physical and experimental conditions. 3D point clouds and RGB images are acquired and processed using an RGBD camera, allowing the robot to autonomously perform robotic assembly tasks without manual reprogramming. In this system, a five-axis hybrid delta robot is intended to perform the assembly operation. In order to detect the workpiece to be assembled by the delta robot, YOLO11 segmentation is applied to the RGB image only. Using the segmentation mask obtained from the model, the point cloud data corresponding to the detected object is extracted from the depth frame. The Iterative Closest Point (ICP) algorithm is then applied to the extracted point cloud to accurately determine the object's position and orientation in three-dimensional space based on the object's initial pose. Thus, it is aimed that the delta robot will be able to perceive the objects around it and gain the competence to continue its task autonomously without the need for human intervention. This approach aims to translate theoretical knowledge into practical applications and to demonstrate the effectiveness of computer vision methods in robotics for solving real-world problems. The developed robotic system aims to contribute to the literature by providing a flexible and innovative infrastructure where artificial intelligence, image processing, and point cloud processing algorithms can be physically tested.