Physical Human-Robot Collaboration using Tactile Information

The aim of the robotics has been, since its beginning, the search of different approaches for supporting humans in complex tasks. Physical human-robot interaction has recently become an indispensable research topic within robotics. Physical Human-Robot Interaction (pHRI) suppose also an important technological advancement by combining improvements in the hardware design of the manipulators and the design of more robust controllers for ensure the safety during the interaction. Assistance using pHRI has been developed usually with force based controllers that generate a motion from the force measurements. The precise intention of the human is difficult to estimate due to the use of generic sensors that are not able to distinguish between external and internal forces of the system. The present thesis shows the development of a robotic controller that assists in a physical human robot interaction. The work is based in the definition of a dexterous system formed by a robot manipulator and a robotic multifingered hand. The main contribution of this set up comes from the use of the multifingered hand which, by using the tactile information collected, permits to differentiate in a more precise way the humans intention behind them. The robot manipulator used is the KUKA LWR4+, a lightweight edundant robotic system used primary in research and development. Furthermore, the multifingered robotic hand chosen provides tactile sensors along both the fingers and the palm. Nowadays, as robotics is present in many different tasks, the use of a flexible griper like the Barrett hand is essential not only for the tactile information itself but also for allowing the grasping of objects with different morphologies. The structure of the thesis starts with an explanatory section of the hardware used and the way it can be controlled. Afterwards, the chapters focus on the practical work follows, initially with experimental interactions between a human user and the manipulator. Consequently, different assistance approaches are analysed on the KUKA robot. Subsequently, the Barrett hand control is described and the tactile information is processed in order to transmit the humans intention. Finally, the implementation and development of a collaborative system is elaborated