TY - JOUR
T1 - A selective muscle fatigue management approach to ergonomic human-robot co-manipulation
AU - Peternel, Luka
AU - Fang, Cheng
AU - Tsagarakis, Nikos
AU - Ajoudani, Arash
N1 - Accepted Author Manuscript
PY - 2019
Y1 - 2019
N2 - In this paper, we propose a method for selective monitoring and management of human muscle fatigue in human-robot co-manipulation scenarios. The proposed approach uses a machine learning technique to learn the complex relationship between individual human muscle forces, arm configuration and arm endpoint force that are provided by a sophisticated offline musculoskeletal model. The estimated muscle forces are used in the fatigue model to estimate the individual muscle fatigue levels online. Two fatigue management protocols are proposed that enable the robot to handle and reduce the human fatigue by altering the configuration of task execution. The first protocol uses optimisation technique to find the optimal position for task execution, where the fatigue-related endurance time can be maximised. The second protocol divides the arm muscles into groups and then alters the direction of endpoint force so that the fatigued muscle group can relax and the relaxed muscle group becomes active. The proposed method has a potential to enable the robot to facilitate safer and more ergonomic working conditions for the human coworker. The main advantage of this approach is that it can operate online, and that all the measurements can be performed by the robot sensory system, which can significantly increase the applicability in real world scenarios. To validate the proposed method, we performed multiple experiments with two collaborative tasks (polishing and drilling) under different conditions.
AB - In this paper, we propose a method for selective monitoring and management of human muscle fatigue in human-robot co-manipulation scenarios. The proposed approach uses a machine learning technique to learn the complex relationship between individual human muscle forces, arm configuration and arm endpoint force that are provided by a sophisticated offline musculoskeletal model. The estimated muscle forces are used in the fatigue model to estimate the individual muscle fatigue levels online. Two fatigue management protocols are proposed that enable the robot to handle and reduce the human fatigue by altering the configuration of task execution. The first protocol uses optimisation technique to find the optimal position for task execution, where the fatigue-related endurance time can be maximised. The second protocol divides the arm muscles into groups and then alters the direction of endpoint force so that the fatigued muscle group can relax and the relaxed muscle group becomes active. The proposed method has a potential to enable the robot to facilitate safer and more ergonomic working conditions for the human coworker. The main advantage of this approach is that it can operate online, and that all the measurements can be performed by the robot sensory system, which can significantly increase the applicability in real world scenarios. To validate the proposed method, we performed multiple experiments with two collaborative tasks (polishing and drilling) under different conditions.
KW - Machine learning
KW - Muscle fatigue
KW - Muscle force estimation
KW - Physical human-robot collaboration
UR - http://www.scopus.com/inward/record.url?scp=85061633513&partnerID=8YFLogxK
U2 - 10.1016/j.rcim.2019.01.013
DO - 10.1016/j.rcim.2019.01.013
M3 - Article
AN - SCOPUS:85061633513
SN - 0736-5845
VL - 58
SP - 69
EP - 79
JO - Robotics and Computer-Integrated Manufacturing
JF - Robotics and Computer-Integrated Manufacturing
ER -