Image processing and ai-based classification of products with six-axis robot arm manufactured by additive manufacturing method

In this thesis, it is aimed that a robot arm with 5+1 degrees of freedom can detect an object in a certain position and shape and provide control accordingly. It is aimed to use the studied theoretical and algorithmic structure in real and simulation applications. Real-time and smart applications have been realized in the application of the robot arm. In our design study, the robot arm control was made by using six stepper motors by making the necessary calculations. The ability of the robot to know the object to which it will go, where it will go, has been gained by image processing technology. First of all, it was designed with robot arm design programs. The designed robot arm was manufactured using PLA material with the additive manufacturing method. A programmable logic controller (PLC), which has a suitable motor and programming card, has been applied to the designed robot arm. Six axes are programmed to the PLC with the interface software. D-H table was calculated according to the limb lengths and axis movements of the robot arm. Transformation matrices were obtained for each axis and forward and inverse kinematics calculations were made. In the study, image processing was used to detect objects and calculate centroids, and U2-Net artificial techniques were used to erase the backgrounds in the images. Color spaces have been converted to HSV color space in order to remove the background on the obtained RGB images and to detect objects by color. As a result of the study, the robot arm, whose parts are produced and whose software is coded, can move on an axis with 5+1 degrees of freedom. In addition, image processing techniques distinguish the color of objects and move to positions at certain points.