Safe human-robot collaboration, especially considering ergonomics, still represents a challenging problem. The objective of this paper is to develop the procedure for assessing human-robot collaboration with the help of computer vision system and electromyography (EMG) data. The conducted study involved three participants performing the polishing task in four different configurations interacting with Franka Emika’s Panda Robot. The robot is controlled in the impedance mode to achieve safe human-robot interaction and, at the same time, to achieve desired robot impedance in the translational and the rotational axis. In order to study human kinematics, the experiment was recorded using four IP cameras, while muscle activity was measured by Trigno Avanti Duo sensors by Delsys to account for human dynamics in the contact task. The collected EMG signals were processed using a bandpass filter, a notch filter, envelope calculation, and normalization. The reconstruction of a human pose from the collected videos was performed using the VIBE algorithm, while body pose parameters (characteristic postural angles) were computed from the output SMPL parametric human model. The obtained results showed significant variations among different body configurations, as verified by the trend of the EMG signals. The proposed approach demonstrates the potential to be an effective tool for enhancing ergonomics in an industrial environment.
Assessment of the Human-Robot Collaborative Polishing Task by Using EMG Sensors and 3D Pose Estimation
Mechan. Machine Science
International Conference on Robotics in Alpe-Adria Danube Region ; 2022 ; Klagenfurt, Austria June 08, 2022 - June 10, 2022
2022-04-23
7 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Incorporating Contextual Knowledge Into Human-Robot Collaborative Task Execution
BASE | 2020
|