Tesi etd-07102024-163127
Link copiato negli appunti
Tipo di tesi
Dottorato
Autore
PENNA, MICHELE FRANCESCO
URN
etd-07102024-163127
Titolo
Development of a portable exoskeleton and adaptive control strategies based on residual movement capabilities to assist and rehabilitate upper-limb impaired individuals
Settore scientifico disciplinare
ING-IND/34
Corso di studi
Istituto di Biorobotica - PHD IN BIOROBOTICA
Commissione
Presidente Prof. VITIELLO, NICOLA
Relatore Prof.ssa CASADIO, MAURA
Relatore Prof.ssa CASADIO, MAURA
Parole chiave
- Wearable robotics
- intention recognition
- rehabilitation robotics
- exoskeletons
Data inizio appello
24/02/2025;
Disponibilità
parziale
Riassunto analitico
Human upper-limb (UL) biomechanics guarantees a sophisticated synergy of motor and sensory capabilities. The occurrence of UL impairments due to traumatic or non traumatic events may affect the execution of common activities of daily living (ADLs), reducing the subject's quality of life. To date, passive orthoses are frequently used to increase the subject's autonomy and foster a social and occupational reintegration. These devices, such as arm slings, splints, or braces, are typically employed to maintain joint alignment and promote increases of the range of motion. However, due to the absence of integrated power sources, the human-device interaction is purely hardware-coded within the passive orthosis structure.
Exoskeletons are mechatronic systems designed to operate in conjunction with the human body. These devices integrate sensory, motor, and cognitive functionalities to provide their users with joint-specific assistance. The employment of these devices has recently attracted significant interest from both research and market domains. Indeed, exoskeletons interact with the human user both from a physical and a cognitive perspective, and can thus be used to observe their user's residual movement capability and amplify it according to adaptive control strategies.
State-of-the-art UL exoskeletons typically integrate advanced mechanical design features, including different actuation strategies, variable transmission joints, and smart physical human-robot interfaces. On the control side, intention decoding algorithms (IDAs) are employed to infer the exoskeleton's user motor intentions and provide assistive torques to amplify them. IDAs can rely on a variety of input signals, such as biological, kinematic and/or kinetic, and environmental information. However, some challenges must be faced to foster the widespread adoption of exoskeletons outside laboratory settings. First, the usability and acceptability of exoskeleton must be enhanced, limiting their weights and encumbrances and improving their customization to better fit the user. Moreover, controllers and IDAs must be designed to provide a task-specific and user-adaptive assistance, especially in the UL framework which typically involves discrete movements of different amplitudes and orientations.
This thesis aims to face the challenges that limit the adoption of UL exoskeletons in everyday life scenarios. The main contribution is twofold. First, the development of a novel portable shoulder-elbow exoskeleton, with the aim of assisting UL impaired individuals. Second, the design, development, and testing of two novel IDAs, exploiting kinematics and electromyographic (EMG) information.
GraCE is a portable and lightweight (<3.5 kg) shoulder-elbow exoskeleton. The exoskeleton has been designed and developed with a user-centered approach, leveraging the results of a focus-group with eleven UL impaired individuals, who suffered brachial plexus injury and stroke. The exoskeleton integrates three main modules: (i) an active elbow module integrating series elastic actuation, (ii) a spring-loaded shoulder module to provide arm gravity compensation torques, and (iii) a spring-loaded scapular module to compensate for reaction forces and track the movements of the scapulo-humeral rhythm.
The control of the GraCE elbow module integrates IDAs to provide adaptive assistance during reaching movements; moreover, the module can be controlled to provide forearm and hand gravity compensation. Two different IDAs were developed and tested on healthy subjects. The first, i.e., "multi degrees-of-freedom adaptive Dynamic Movement Primitive" (Multi-DOF aDMP) algorithm, is an IDA aiming to assist users with UL impairments who can initiate the movements by themselves. The algorithm relies on the observation of the initial part of the user's movement through joint angle measures and it assists the user's hand movement during reaching tasks, estimating in real-time the phase of the movement and the final position of the hand. The second, i.e., muscle synergies-based algorithm (Syn-ID), is an IDA employing EMG signals to detect the movement onset and infer the direction of the movement during reaching tasks. The algorithm uses Gaussian Mixture Model and a probability accumulation-based logic. The algorithm aims to train UL motor functions in patients with limited movement capabilities, who can express muscle activities but need an amplification to drive the arm towards the desired direction. Before being tested on GraCE, the algorithms were tested on the NESM-gamma shoulder-elbow exoskeleton, during planar reaching movements tasks.
In conclusion, this thesis aims to mark a step further for the adoption of wearable robots in everyday-life practice. The presented results pave the way to the development of novel strategies facilitating the seamless adaptation of the exoskeleton to the user's need.
Exoskeletons are mechatronic systems designed to operate in conjunction with the human body. These devices integrate sensory, motor, and cognitive functionalities to provide their users with joint-specific assistance. The employment of these devices has recently attracted significant interest from both research and market domains. Indeed, exoskeletons interact with the human user both from a physical and a cognitive perspective, and can thus be used to observe their user's residual movement capability and amplify it according to adaptive control strategies.
State-of-the-art UL exoskeletons typically integrate advanced mechanical design features, including different actuation strategies, variable transmission joints, and smart physical human-robot interfaces. On the control side, intention decoding algorithms (IDAs) are employed to infer the exoskeleton's user motor intentions and provide assistive torques to amplify them. IDAs can rely on a variety of input signals, such as biological, kinematic and/or kinetic, and environmental information. However, some challenges must be faced to foster the widespread adoption of exoskeletons outside laboratory settings. First, the usability and acceptability of exoskeleton must be enhanced, limiting their weights and encumbrances and improving their customization to better fit the user. Moreover, controllers and IDAs must be designed to provide a task-specific and user-adaptive assistance, especially in the UL framework which typically involves discrete movements of different amplitudes and orientations.
This thesis aims to face the challenges that limit the adoption of UL exoskeletons in everyday life scenarios. The main contribution is twofold. First, the development of a novel portable shoulder-elbow exoskeleton, with the aim of assisting UL impaired individuals. Second, the design, development, and testing of two novel IDAs, exploiting kinematics and electromyographic (EMG) information.
GraCE is a portable and lightweight (<3.5 kg) shoulder-elbow exoskeleton. The exoskeleton has been designed and developed with a user-centered approach, leveraging the results of a focus-group with eleven UL impaired individuals, who suffered brachial plexus injury and stroke. The exoskeleton integrates three main modules: (i) an active elbow module integrating series elastic actuation, (ii) a spring-loaded shoulder module to provide arm gravity compensation torques, and (iii) a spring-loaded scapular module to compensate for reaction forces and track the movements of the scapulo-humeral rhythm.
The control of the GraCE elbow module integrates IDAs to provide adaptive assistance during reaching movements; moreover, the module can be controlled to provide forearm and hand gravity compensation. Two different IDAs were developed and tested on healthy subjects. The first, i.e., "multi degrees-of-freedom adaptive Dynamic Movement Primitive" (Multi-DOF aDMP) algorithm, is an IDA aiming to assist users with UL impairments who can initiate the movements by themselves. The algorithm relies on the observation of the initial part of the user's movement through joint angle measures and it assists the user's hand movement during reaching tasks, estimating in real-time the phase of the movement and the final position of the hand. The second, i.e., muscle synergies-based algorithm (Syn-ID), is an IDA employing EMG signals to detect the movement onset and infer the direction of the movement during reaching tasks. The algorithm uses Gaussian Mixture Model and a probability accumulation-based logic. The algorithm aims to train UL motor functions in patients with limited movement capabilities, who can express muscle activities but need an amplification to drive the arm towards the desired direction. Before being tested on GraCE, the algorithms were tested on the NESM-gamma shoulder-elbow exoskeleton, during planar reaching movements tasks.
In conclusion, this thesis aims to mark a step further for the adoption of wearable robots in everyday-life practice. The presented results pave the way to the development of novel strategies facilitating the seamless adaptation of the exoskeleton to the user's need.
File
| Nome file | Dimensione |
|---|---|
Ci sono 1 file riservati su richiesta dell'autore. |
|