The introduction of an active exoskeleton that enhances the operator power in the manufacturing field was demonstrated in literature to lead to beneficial effects in terms of reducing fatiguing and the occurrence of musculoskeletal diseases. However, a large number of manufacturing operations would not benefit from power increases because it rather requires the modulation of the operator stiffness. However, in literature, considerably less attention was given to those robotic devices that regulate their stiffness based on the operator stiffness, even if their introduction in the line would aid the operator during different manipulations respect with the exoskeletons with variable power. In this thesis the description of the command logic of an exoskeleton for manufacturing applications, whose stiffness is modulated based on the operator stiffness, is described. Since the operator stiffness cannot be mechanically measured without deflecting the limb, an estimation based on the superficial Electromyographic signal is required. A model composed of 1 joint and 2 antagonist muscles was developed to approximate the elbow and the wrist joints. Each muscle was approximated as the Hill model and the analysis of the joint stiffness, at different joint angle and muscle activations, was performed. The same Hill muscle model was then implemented in a 2 joint and 6 muscles (2J6M) model which approximated the elbowshoulder system. Since the estimation of the exerted stiffness with a 2J6M model would be quite onerous in terms of processing time, the estimation of the operator endpoint stiffness in realtime would therefore be questionable. Then, a linear relation between the endpoint stiffness and the component of muscle activation that does not generate any endpoint force, is proposed. Once the stiffness the operator exerts was estimated, three command logics that identifies the stiffness the exoskeleton is required to exert are proposed. These proposed command logics are: Proportional, Integral 1 s, and Integral 2 s. The stiffening exerted by a device in which a Proportional logic is implemented is proportional, sample by sample, to the estimated stiffness exerted by the operator. The stiffening exerted by the exoskeleton in which an Integral logic is implemented is proportional to the stiffness exerted by the operator, averaged along the previous 1 second (Integral 1 s) or 2 seconds (Integral 2 s). The most effective command logic, among the proposed ones, was identified with empirical tests conducted on subjects using a wrist haptic device (the Hi5, developed by the Bioengineering group of the Imperial College of London). The experimental protocol consisted in a wrist flexion/extension tracking task with an external perturbation, alternated with isometric force exertion for the estimation of the occurrence of the fatigue. The fatigue perceived by the subject, the tracking error, defined as the RMS of the difference between wrist and target angles, and the energy consumption, defined as the sum of the squared signals recorded from two antagonist muscles, indicated the Integral 1 s logic to be the most effective for controlling the exoskeleton. A logistic relation between the stiffness exerted by the subject and the stiffness exerted by the robotic devices was selected, because it assured a smooth transition between the maximum and the minimum stiffness the device is required to exert. However, the logistic relation parameters are subjectspecific, therefore an experimental estimation is required. An example was provided. Finally, the literature about variable stiffness actuators was analyzed to identify the most suitable device for exoskeleton stiffness modulation. This actuator is intended to be integrated on an existing exoskeleton that already enhances the operator power based on the operator Electromyographic signal. The identified variable stiffness actuator is the DLR FSJ, which controls its stiffness modulating the preload of a single spring.
Concept of an exoskeleton for industrial applications with modulated impedance based on Electromyographic signal recorded from the operator
BORZELLI, DANIELE
2019
Abstract
The introduction of an active exoskeleton that enhances the operator power in the manufacturing field was demonstrated in literature to lead to beneficial effects in terms of reducing fatiguing and the occurrence of musculoskeletal diseases. However, a large number of manufacturing operations would not benefit from power increases because it rather requires the modulation of the operator stiffness. However, in literature, considerably less attention was given to those robotic devices that regulate their stiffness based on the operator stiffness, even if their introduction in the line would aid the operator during different manipulations respect with the exoskeletons with variable power. In this thesis the description of the command logic of an exoskeleton for manufacturing applications, whose stiffness is modulated based on the operator stiffness, is described. Since the operator stiffness cannot be mechanically measured without deflecting the limb, an estimation based on the superficial Electromyographic signal is required. A model composed of 1 joint and 2 antagonist muscles was developed to approximate the elbow and the wrist joints. Each muscle was approximated as the Hill model and the analysis of the joint stiffness, at different joint angle and muscle activations, was performed. The same Hill muscle model was then implemented in a 2 joint and 6 muscles (2J6M) model which approximated the elbowshoulder system. Since the estimation of the exerted stiffness with a 2J6M model would be quite onerous in terms of processing time, the estimation of the operator endpoint stiffness in realtime would therefore be questionable. Then, a linear relation between the endpoint stiffness and the component of muscle activation that does not generate any endpoint force, is proposed. Once the stiffness the operator exerts was estimated, three command logics that identifies the stiffness the exoskeleton is required to exert are proposed. These proposed command logics are: Proportional, Integral 1 s, and Integral 2 s. The stiffening exerted by a device in which a Proportional logic is implemented is proportional, sample by sample, to the estimated stiffness exerted by the operator. The stiffening exerted by the exoskeleton in which an Integral logic is implemented is proportional to the stiffness exerted by the operator, averaged along the previous 1 second (Integral 1 s) or 2 seconds (Integral 2 s). The most effective command logic, among the proposed ones, was identified with empirical tests conducted on subjects using a wrist haptic device (the Hi5, developed by the Bioengineering group of the Imperial College of London). The experimental protocol consisted in a wrist flexion/extension tracking task with an external perturbation, alternated with isometric force exertion for the estimation of the occurrence of the fatigue. The fatigue perceived by the subject, the tracking error, defined as the RMS of the difference between wrist and target angles, and the energy consumption, defined as the sum of the squared signals recorded from two antagonist muscles, indicated the Integral 1 s logic to be the most effective for controlling the exoskeleton. A logistic relation between the stiffness exerted by the subject and the stiffness exerted by the robotic devices was selected, because it assured a smooth transition between the maximum and the minimum stiffness the device is required to exert. However, the logistic relation parameters are subjectspecific, therefore an experimental estimation is required. An example was provided. Finally, the literature about variable stiffness actuators was analyzed to identify the most suitable device for exoskeleton stiffness modulation. This actuator is intended to be integrated on an existing exoskeleton that already enhances the operator power based on the operator Electromyographic signal. The identified variable stiffness actuator is the DLR FSJ, which controls its stiffness modulating the preload of a single spring.File  Dimensione  Formato  

Tesi discussa per il conseguimento del titolo di dottore di ricerca in Ingegneria Meccanica svolta presso il corso di dottorato in ingegneria meccanica del Politecnico di Torino.pdf
accesso aperto
Dimensione
5.93 MB
Formato
Adobe PDF

5.93 MB  Adobe PDF  Visualizza/Apri 
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/65393
URN:NBN:IT:POLITO65393