Dual-Arm Robot Cooperation and Path Planning for Picking, Handing-Over, and Placing

This thesis investigates the cooperation and path planning strategies for dual-arm robots in tasks involving picking, handing over, and placing objects. The primary focus is on developing a compact and comprehensive algorithm that subdivides each move into several sub-moves to enhance control of the robots through ROS. Additionally, it links the path planner environment to the simulation environment for more accurate and near-real-time simulations. While a solid rectangular cube is used as the manipulated object, future developments could involve deformable objects. The simulations were conducted using Nvidia Omniverse Isaac Sim, a powerful software that allows for realistic simulations influenced by physical features such as gravity and friction. This simulation environment is connected to the path planner, MoveIt, via the ROS Bridge and Omnigraph, Isaac Sim’s action graph. This connection enables synchronization of the robots’ movements in both environments. Moreover, the system’s behavior is made more interactive by linking the status of the manipulated object in both environments. The position and orientation of the object are published from the simulation environment (Isaac Sim) through ROS and utilized by the path planner (MoveIt) to initialize or finalize the pick and place tasks. Several aspects can be improved in future work, such as enabling simultaneous arm movements, creating more complex and cluttered environments, or integrating more than two robotic arms for similar tasks. These improvements offer a basis for further research based on this project.