Improving Human-Machine Interaction
This thesis studies human and machine interaction. For better interaction between humans and machines, this thesis aims to address three issues that remain unanswered in literature. Three objectives are proposed in this thesis to address the three issues, and the objectives are: (i) identification of the core capabilities of a Human Assistance System (HAS) and study of implementation strategy of the core capabilities; (ii) development of a framework for improving the accuracy of human mind state inference; (iii) study of the effect of representation of the machine’s state (which is represented in a “natural” way) on the user’s actions. By a natural way, it is meant a way that contains emotions known to be always present in humans (or human emotions in short). The study includes theoretical development, experimentation, and prototype implementation. This thesis has concluded: (1) the core capabilities to be addressed in designing a HAS are transparency, communication, rationale, cognition and task-sharing and they can be implemented with the existing technologies including fuzzy logics, Petri Net and ACT-R (Adaptive Control of Thought-Rational); (2) expert opinion elicitation technique is a promising method to construct a more general framework for integrating various algorithms on human state inference; (3) there is a significant effect of the representation of the machine’s state on the user’s actions. The main contributions of this thesis are: (1) provision of a case study for the proof-of-concept of HAS in the area of Computer Aided Design (CAD); (2) provision of an integrated framework for fatigue inference for improved accuracy, being readily generalized to inference of other mind states; (3) generation of a new knowledge regarding the effect of the natural representation of a machine’s states on the user’s actions. These contributions are significant in human-machine science and technology. The first contribution may lead to the development of a new generation CAD system in the near future. The second contribution provides a much powerful technology for human mind inference, which is a key capability in HAS, and the third contribution enriches the science of human-machine interaction and will give impact to the field of Artificial Intelligence (AI) as well. The application of the result of this thesis is rehabilitation, machine learning, etc.
DegreeMaster of Science (M.Sc.)
SupervisorZhang, W. J. (Chris)
CommitteeBurton, Richard; Chen, Daniel
Copyright DateAugust 2011