Abstract
The ability to perform motor activities in interaction with the physical environment, such as walking over uneven terrains or balancing on a moving train, is related to the capacity to regulate the mechanical properties of our joints, quantified by human joint impedance. Typically, to experimentally identify joint impedance, data-driven modeling techniques are applied: a position perturbation (input) is applied to a joint, and an impedance model is dynamically fit between the perturbation and the overall response torque around the joint (output). Data-driven modeling of the lower limb’s joint impedance can provide important information for the design of wearable bionic devices and guide the rationale for robust and transparent control.
The first contribution of this thesis is to develop and evaluate two frequency based data-driven methodologies applicable to model time-varying human joint impedance during out-of-the-laboratory conditions. The methodologies are described theoretically, evaluated in realistic simulations, and applied to analyzing human experimental data relevant to wearable devices.
The second contribution is to illustrate future pathways for applying data-driven
techniques in combination with biomechanical models to improve the understanding of physiological processes contributing to joint impedance.
Finally, reflections on data-driven methodologies for better modeling human
biomechanical behavior and more intuitively interacting with it are provided,
highlighting limitations and considerations to best use these powerful tools.
The first contribution of this thesis is to develop and evaluate two frequency based data-driven methodologies applicable to model time-varying human joint impedance during out-of-the-laboratory conditions. The methodologies are described theoretically, evaluated in realistic simulations, and applied to analyzing human experimental data relevant to wearable devices.
The second contribution is to illustrate future pathways for applying data-driven
techniques in combination with biomechanical models to improve the understanding of physiological processes contributing to joint impedance.
Finally, reflections on data-driven methodologies for better modeling human
biomechanical behavior and more intuitively interacting with it are provided,
highlighting limitations and considerations to best use these powerful tools.
| Original language | English |
|---|---|
| Awarding Institution |
|
| Supervisors/Advisors |
|
| Award date | 2 Mar 2023 |
| Publication status | Published - 2023 |