<![CDATA[Technology related to robotics has developed rapidly in recent years. In manufacturing production lines, an industrial pick-and-place robot is used to efficiently move objects from one location to another. In most approaches, this robot automates the repetitive task from one exact start position. However, the task of collecting objects from various positions in the robot workspace still introduces challenges in terms of object positional detection and movement accuracy. In this paper, an arm robot system equipped with automatic color-based object recognition and position control was proposed. The robot was able to detect multiple target object positions automatically without any need to plan a fixed movement beforehand. In the construction of the experiment platform, a Pixy2 camera sensor with color recognition ability was integrated into a 4-DoF Dobot Magician arm robot. Furthermore, a coordinate transformation was derived and implemented to achieve an accurate positional robot movement. The coordinate transformation performed a mapping from the Camera Coordinate System (CCS), which was initialized from image pixel values to the Robot Coordinate System (RCS), which was finalized to the robot’s actuator input signals. Prior to the implementation, the robot underwent a color calibration and position calibration. Thereafter, a set of color signatures was obtained and any object position in the camera’s field of view can be matched with any end-effector position in the robot’s workspace. Three experiment setups were conducted to evaluate the proposed system. Limited to one lighting condition, the robot was commanded to pick-and-place objects based on the criteria of all 3 colors, 1 specific color, and 2 specific colors. The robot performed perfectly to pick and place the objects, achieving a 100% success rate in terms of object color detection and pick-and-place. The positive results encouraged further investigation in different actuator actions and greater work areas.]]>
