ROBOT CONTROLLING APPLICATION TO EFFICIENTLY ARRANGE THE OBJECTS ON THE TABLE AND STACK OBJECTS IN THE RIGHT SEQUENCE
The main focus of the study was to maintain the tidiness of a workspace. This study identified tidiness as a prominent factor in human life. But conventional practices involve more time and a considerable amount of human labor and it is hard to maintain the tidiness. So, this study was conducted believing that the robotic manipulators and AI (artificial intelligence) become handy in this scenario. Hence, this study proposed and implemented the robot controlling application to efficiently arrange the objects on a table and stack objects in the right sequence as a proof of concept.
This application was able to detect the object on the table, identify them and was capable of placing them at a given location. This application contains three main components namely, robot arm, power supply, camera and microcontroller. The object detection component used its algorithm consisting of image processing capabilities to detect objects on the table using a camera. Then, the coordinate conversion component was able to convert the image pixel coordinate into real world coordinates. Further, the object identification component used a trained machine learning model to identify the detected and located objects. Destination coordinates can be decided as per the identification of the object. After providing the present coordinates and the destination coordinates, the robot arm was capable of picking and placing the object in the correct place using the inverse kinematics. This robot arm consisted of four servo motors.
 |
| Figure 1 : Robot arm |
The object detection algorithm gave 94% positive results and the object identification component gave 80% accuracy of positive identification results. Further, this arm was capable of picking and placing an object on the table in less than one minute. Future modifications of this system can be stated as, path and time optimization, identification of the several objects and movable robot arm. Moreover, this concept can be effectively used in cargo management applications in dockyards.
 |
| Figure 2 : Components of the Robot arm |
 |
| Figure 3: Identification results |
Supervisors:
Dr. S. Fernando
Dr. P. Kumarage