Motion planning of robotic manipulation: developing a framework for efficient path planning and obstacle avoidance

Abstract

This study investigates the potential of using ROS2 as a possible methodology to develop a dynamic system, capable of controlling pick-and-place operations with a moving conveyor belt, which is an autonomous and repeatable task. Throughout the paper, a detailed methodology and implementation of the system showed the integration of different hardware, interfaced using various ROS2 packages to form a cohesive system to perform the tasks. Various tests were conducted on each aspect of the system, showcasing demonstrations of the robotic arm performing tasks such as static and dynamic grasping. Each component was examined and tested in detail and its capabilities, limitations and results were discussed. By using computer vision, data acquired from the camera was processed and sent to the ROS2 controller to plan a path for the robotic arm to successfully grasp various common objects at different positions and orientations. This project was also integrated with another Final Year Project that explores into the potential of utilising deep learning algorithms for object detection for grasping, and the results of the project integration were discussed. Finally, this study concludes with proposals for future improvements.