| Resumo: | Human Robot Collaboration has become a prominent subject in industry settings over the years. The insurgence of collaborative robots and the advantages they bring to assembly lines have caused the need to exploit their hardware and create new systems that can enable the safe execution of collaborative tasks between humans and robots. Such advantages include their reduced size and cost, their lightweight construction, their force/precision assistance hardware and their safety requirements, that must be met in order to be considered collaborative. In this Dissertation, we propose a set of tools and techniques that enable the execution of collaborative tasks between a human and a robotic manipulator. Such techniques include dynamic payload compensation at the end effector of the robot, identification of moving obstacles in the environment, and the creation of high level task plans. This techniques can then be joined together to create various collaborative tasks. Examples of such, both implemented and tested, are the transfer of objects between human and robot, precise hand guiding of the robot at the end effector, hand guided manipulation of dynamically coupled objects, and the execution of an industrial task with collision avoidance. All of this tasks were grouped in a global state machine that allows the user to seamlessly interact with the robot, change the task in execution while receiving constant feedback. These tasks were implemented and tested in a real setting with an Universal Robots UR10e. The setup consisted on a shared workspace where the human and the robot could physically interact and perform the proposed tasks.
|
|---|