Tesi etd-10132023-225301
Link copiato negli appunti
Tipo di tesi
Corso Ordinario Secondo Livello
Autore
AMATO, LORENZO
URN
etd-10132023-225301
Titolo
Whole body control of a 6-DOF lower-limb exoskeleton
Struttura
Cl. Sc. Sperimentali - Ingegneria
Corso di studi
INGEGNERIA - INGEGNERIA
Commissione
Tutor Prof.ssa CARROZZA, MARIA CHIARA
Relatore Prof. VITIELLO, NICOLA
Membro Dott.ssa CREA, SIMONA
Membro Dott. TRIGILI, EMILIO
Presidente Prof. CIARAMELLA, ERNESTO
Membro Prof. ODDO, CALOGERO MARIA
Membro Prof. SOLAZZI, MASSIMILIANO
Membro Prof. SABATINI, ANGELO MARIA
Membro Prof. FONTANA, MARCO
Membro Prof.ssa niero, monia
Membro Prof. CIPRIANI, CHRISTIAN
Relatore Prof. VITIELLO, NICOLA
Membro Dott.ssa CREA, SIMONA
Membro Dott. TRIGILI, EMILIO
Presidente Prof. CIARAMELLA, ERNESTO
Membro Prof. ODDO, CALOGERO MARIA
Membro Prof. SOLAZZI, MASSIMILIANO
Membro Prof. SABATINI, ANGELO MARIA
Membro Prof. FONTANA, MARCO
Membro Prof.ssa niero, monia
Membro Prof. CIPRIANI, CHRISTIAN
Parole chiave
- dynamic modelling
- human-robot interaction.
- lower limb exoskeletons
- rehabilitation robotics
- whole body control
Data inizio appello
15/12/2023;
Disponibilità
parziale
Riassunto analitico
Stroke and traumatic spinal cord injury are leading causes of lower-limb impairments. Gait recovery is a primary goal in the rehabilitation of these patients. Currently, gait rehabilitation primarily relies on physical therapy, where the physiotherapist supports the correct patient’s movement patterns via physical assistance These methods can place a significant physical strain on therapists, potentially leading over time to work-related injuries. Robotic rehabilitation, especially exoskeletons, offer a solution to these challenges. However, for effective integration of these devices in a rehabilitation setting, it is essential that exoskeletons offer precise torque control without hindering the patient's movements.
To enhance torque control performance and robot transparency, this work presents a whole-body control system for a full lower-limb exoskeleton with six active degrees of freedom. This control system compensates for robot dynamics, accounting for factors such as exoskeleton gravity and joint friction. Gravity compensation is achieved by modeling the exoskeleton as a seven-link mechanism with six active joints. The ground of the mechanism is adapted considering the gait states. A state machine was implemented identifying the gait states using the foot pressure sensors. The friction was experimentally identified, and a linear friction model for friction compensation was implemented. Additionally, the controller allows the definition of desired interaction torque. Two different strategies were implemented, transparent control where the user is free to move, and the interaction module. This last module was realized with frequency domain designed virtual stiffness, formulated as a one-pole one-zero transfer function, with behavior defined by the values of gains. This formulation mitigates the high frequency component of the tracking error, which are the main cause of instability in impedance controls.
The whole-body controller was tested and evaluated on a bench test setup and with a healthy subject walking on a treadmill at 1,1 km/h for 3 minutes. Results showed that the controller, during walking, achieved normalized interaction torques comparable to those of other controllers proposed in the literature. The dynamic compensation strategy reduced interaction torque and improved the subject's range of motion. The frequency dependent stiffness was tested on a bench test setup and with a healthy subject walking on a treadmill at different speeds and it was compared with the classic impedance control implementation. The frequency dependent stiffness was found to effectively emphasize the reference signal’s low-frequency components and to provide greater stability when subjected to high-frequency reference trajectories.
To enhance torque control performance and robot transparency, this work presents a whole-body control system for a full lower-limb exoskeleton with six active degrees of freedom. This control system compensates for robot dynamics, accounting for factors such as exoskeleton gravity and joint friction. Gravity compensation is achieved by modeling the exoskeleton as a seven-link mechanism with six active joints. The ground of the mechanism is adapted considering the gait states. A state machine was implemented identifying the gait states using the foot pressure sensors. The friction was experimentally identified, and a linear friction model for friction compensation was implemented. Additionally, the controller allows the definition of desired interaction torque. Two different strategies were implemented, transparent control where the user is free to move, and the interaction module. This last module was realized with frequency domain designed virtual stiffness, formulated as a one-pole one-zero transfer function, with behavior defined by the values of gains. This formulation mitigates the high frequency component of the tracking error, which are the main cause of instability in impedance controls.
The whole-body controller was tested and evaluated on a bench test setup and with a healthy subject walking on a treadmill at 1,1 km/h for 3 minutes. Results showed that the controller, during walking, achieved normalized interaction torques comparable to those of other controllers proposed in the literature. The dynamic compensation strategy reduced interaction torque and improved the subject's range of motion. The frequency dependent stiffness was tested on a bench test setup and with a healthy subject walking on a treadmill at different speeds and it was compared with the classic impedance control implementation. The frequency dependent stiffness was found to effectively emphasize the reference signal’s low-frequency components and to provide greater stability when subjected to high-frequency reference trajectories.
File
| Nome file | Dimensione |
|---|---|
Ci sono 1 file riservati su richiesta dell'autore. |
|