An Intuitive Android Interface for Robot Arm Motion Control

Sammanfattning

This project focuses on developing an intuitive and responsive Android interface to facilitate motion planning for a robot arm equipped with a singular RGB-D Camera. It leverages Robot Operating System 2 (ROS2) and MoveIt2 for back-end frameworks and Kotlin for the front-end, communicating via a Transfer Control Protocol (TCP) python server. The interface provides two waypoint generation modes: a touch based Free Drawing Mode and an AI Mode, to ensure smooth movements along the surface with the gripper maintaining a perpendicular orientation. It also features real-time video and data feedback during operation. This project focused mainly on gas leak detection, but can be adapted for various other high precision surface treatment tasks. The secondary objective is to explore an AI-assisted 2D background removal method to bypass the need for cloud subtraction, thus making the system plug and play. The system performed well in a simulated environment, but faced certain limitations. Waypoints are generated from a single camera 2D perspective, which necessitates object rotation. Additionally, the background removal method may struggle under sub-optimal conditions such as when the object is hard to disginguish from the background. A small survey was conducted (n=10), where non-expert users rated the interface 4.25/5 for ease of use, with the Free Drawing Mode receiving 3.8/5 and the AI Mode 4.9/5. In summary, this project contributes to the field of human-robot-interaction by offering a user-friendly Android interface for high-precision industrial robotic arm tasks, while also investigating the role of AI to simplify the set up process.