Development of a teleoperated hand-arm robotic platform for the evaluation of shared autonomy algorithms

Humans are unable to operate in hazardous or inaccessible environments. However, the need to perform specific tasks in those places and the curiosity to discover unexplored environments have increasingly turned the spotlight on teleoperation. A teleoperation system requires the presence of a remote robot controlled by a human operator. Teleoperation plays a significant role in different areas, like the nuclear industry, mining, underwater exploration, surgery, etc. For difficult tasks, it might be easier for the operators to delegate some actions to the robot, thus sharing the autonomy of the task execution with the robot. The present master's thesis aims at the development of a teleoperation system for the evaluation of shared autonomy algorithms. The teleoperated system comprises a 6-axis industrial robot (Universal Robot, model UR5) and an anthropomorphic artificial hand (Prensilia SRL, model MIA). The teleoperation system is composed of a wearable inertial motion capture system (Perception Neuron, model PN 32 V2 ) and a dataglove (CyberGlove, model CyberGlove II) which allow measuring the motion of the operator’s arm and hand, respectively. The software has been implemented in ROS (Robot Operating System), which is a flexible framework for writing robot software, using the C++ programming language. For this application, different teleoperation strategies have been implemented for the hand and the arm. The dataglove is used to classify the hand closing and opening actions, which are then automatically executed by the artificial hand. The developed decoding algorithm has been tested with different subjects, which were asked to open and close their hand while moving the arm after a short calibration procedure, without errors. The delay introduced by this algorithm is negligible. The pose of the palm, retrieved by the measurements of the inertial motion capture system, is used to proportionally control the Tool Center Point (TCP) of the robotic arm. An experiment has been carried out to evaluate the performances of the system. Participants were asked to wear the inertial motion capture system and place their right hand on a holder attached to a collaborative robot arm (Universal Robot, model UR5). This robot was programmed to execute precise motions that were then compared to the positions and orientations computed by the algorithm and executed by the teleoperated robot arm. The performances have been evaluated in terms of precision, accuracy, delay and drift. Finally, a qualitative evaluation of the whole system has been carried out on an industrial use case, showing promising capabilities also for teaching the robot to execute human-inspired tasks.