Mechanical Engineering - Theses
Permanent URI for this collection
Search Results
1 - 2 of 2
-
ItemHuman learning in visuomotor field: modeling and experimental studies( 2015)As the saying “practice makes perfect” goes, humans have the ability to learn from experiences. Understanding how humans learn or re-learn tasks/skills have drawn lot attentions from evolutionary biology to cognitive science to neurobiology. Moreover, it would play an important role in fully exploring the learning ability to recover the human motor functions for stroke patients, leading to better or more efficient therapy plans and novel apparatus to aid therapists in recovering patients. As learning may be viewed as a process, concepts of “control” and “feedback” have been successfully used to provide reasonable explanations of learning behaviours observed in many carefully designed experiments for healthy subjects. This work attempts to understand whether humans use a common strategy to learn a task when provided with different environments. To do this, this work aims at 1) Designing human-involved experiments to find out whether humans exhibit different learning behaviours in the same task and environment by different training strategies; 2) Building a computational model to capture the major characteristic of human learning observed from the experiments. The experimental results have shown that under different training strategies, humans will respond differently for the same task and environment, indicating that the training strategies play a crucial role in learning tasks and skills. Quantitative computational model based on novel iterative learning control laws are constructed to propose the underlying mechanism of the different strategies and to provide reasoning for the use of such strategies by humans.
-
ItemHuman motor control: simulation and system identification( 2010)It has been found that humans are able to interact with their environment by adjusting their musculoskeletal system properties. This trait allows the human to perform many different tasks ranging from opening doors to wood carving. Current computational models based on the Electromyography (EMG) Force Control Hypothesis in the literature have difficulties in explaining various ideas of human motor control such as the presence of equifinality and the presence of postures in movement. These ideas have been explained previously using the Equilibrium Point Hypothesis (EPH). However, there exist few rigorous mathematical frameworks which account for the ideas presented in the EPH, resulting in few computational models of human motor control based on the EPH. This work proposes a computational framework to simulate human motor control using the Equilibrium Point Hypothesis. The computational framework is constructed using the ideas of Operational Space Formulation which can account for the presence of posture in movement. Furthermore, the framework introduces the idea of Equilibrium Model which, when used in the context of Model Reference Adaptive Control and Iterative Learning Control, is able to explain human motor control and learning based on the idea of equifinality. The computational framework is applied to construct a model which simulates human subjects performing the task of reaching for a target in different dynamic environments generated by a robot. The results of the simulation are compared qualitatively and quantitatively with the experimental data collected from human subjects.