by Hon Nin Chow. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (leaves 79-84). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Motivation --- p.1 / Chapter 1.2 --- Organization of the Thesis --- p.3 / Chapter 2 --- Literature Survey --- p.6 / Chapter 2.1 --- Learning-by-Demonstration --- p.6 / Chapter 2.2 --- Neural Networks --- p.7 / Chapter 2.3 --- Navigation Learning --- p.8 / Chapter 2.4 --- Localization --- p.9 / Chapter 2.5 --- Robotic Wheelchair --- p.10 / Chapter 3 --- System Implementation --- p.12 / Chapter 3.1 --- Hardware Platform --- p.12 / Chapter 3.2 --- Software Platform --- p.14 / Chapter 3.3 --- Basic Functionality --- p.15 / Chapter 3.3.1 --- Collision Avoidance --- p.15 / Chapter 3.3.2 --- Wearable Eye-jaw Control Interface --- p.16 / Chapter 4 --- Learning Human Navigational Skill --- p.22 / Chapter 4.1 --- Introduction --- p.22 / Chapter 4.2 --- Problem Formulation --- p.23 / Chapter 4.3 --- Approach --- p.23 / Chapter 4.4 --- Experimental Study --- p.26 / Chapter 4.4.1 --- Settings --- p.26 / Chapter 4.4.2 --- Results --- p.30 / Chapter 4.5 --- Discussions --- p.31 / Chapter 5 --- Learning from Multi-phase Demonstrations --- p.33 / Chapter 5.1 --- Introduction --- p.33 / Chapter 5.2 --- Problem Formulation --- p.34 / Chapter 5.3 --- Approach --- p.35 / Chapter 5.4 --- Experimental Study --- p.35 / Chapter 5.4.1 --- Settings --- p.35 / Chapter 5.4.2 --- Results --- p.37 / Chapter 5.5 --- Evaluation of Learning Performance --- p.37 / Chapter 5.6 --- Discussions --- p.43 / Chapter 6 --- Localization Learning --- p.44 / Chapter 6.1 --- Introduction --- p.44 / Chapter 6.2 --- Problem Formulation --- p.45 / Chapter 6.3 --- Approach --- p.45 / Chapter 6.4 --- Experimental Study --- p.46 / Chapter 6.4.1 --- Settings --- p.46 / Chapter 6.4.2 --- Result 1: Localization Performance --- p.47 / Chapter 6.4.3 --- Result 2: Similar Sensor Patterns in Various Configurations . --- p.53 / Chapter 6.4.4 --- Result 3: Small Variations in Major Dimensions of Environ- mental Feature along the Route --- p.53 / Chapter 6.5 --- Discussions --- p.59 / Chapter 6.5.1 --- Accuracy --- p.59 / Chapter 6.5.2 --- Choices of Sensor-Configuration Mappings --- p.60 / Chapter 7 --- Conclusion --- p.62 / Chapter 7.1 --- Contributions --- p.62 / Chapter 7.2 --- Future Work --- p.65 / Chapter A --- Cascade Neural Network --- p.67 / Chapter B --- Trajectories for the Navigation Learning in Chapter 4 --- p.69 / Chapter C --- Publications Resulted from the Study --- p.78

M. Tech. Electrical Engineering. / Modelizes, simulate and analyse the behaviour of an ordinary manual propelled wheelchair (MPW) when equipped with a mechatronic drive module. This serves as a preliminary study towards investigating whether a suitable mechatronic drive module could be designed and easily plugged on an ordinary MPW without any difficulties to obtain full propulsion of the chair with the use of a joystick for navigation purposes. For modelling purposes, a dynamic systems modelling method called Bond Graph was used.

M. Tech. Electrical Engineering. / Wheeled Mobile Platforms have considerable acceptance and dominance in the field of transportation. While most of these platforms have been designed for able users, very few special designs are available for the physically disabled persons. With increasing numbers of disabled peoples around the world, accommodative and safer displacement means are called for. A normal wheelchair platform with two front castor wheel and two independently driven rear wheels is therefore considered in this study for modelling and control. Based on the nature of persons using wheelchairs, better and much easy to control wheelchair platforms are necessitated. To achieve such objective, this research is focused on two main threads: dynamic modelling and control. In modelling, the aim is to present a dynamic model of wheelchair platform that takes into account slipping parameters and frictional/traction forces experienced on motion. It also intends to account for the effects of gravitational forces that would be experienced by the wheelchair and its effect on the platforms during uphill and downhill movements. Lagrange formalism is utilised in the design of this dynamic model. In control, the objective is to ensure that the platform tracks the reference linear velocity and the reference angular orientation as desired. With such control commands, a simple access device may be employed to ensure that people with severe disability also have a chance of controlling the wheelchair. Since no zero dynamics arise with linear velocity and angular orientation as the platforms outputs, standard input-output feedback linearisation is considered and applied in the linearisation of the model and in the development of the control law. Successfully simulated results demonstrating the performance of the proposed dynamic model and control law are presented for verification. The entire dynamic model and the controller are simulated in a software tool MATLAB and SIMULINK.

In the wheelchair family, a manual wheelchair is most commonly used by the lower-limb disabled, while its manual mode creates the lowest mechanical effi-ciency and the highest rate of upper-limb injury. Electric wheelchairs on the market are much more labor-saved; however, with their drive systems mostly combining with the bodies of the wheelchair, electric wheelchairs appear to be bulky and are not easily-carried. While in a limited space, on a short-distance moving, or in/off the elevator, electric wheelchairs cannot be practicably switched to manual ones. Thus, this text aims to develop a switchable dou-ble-action manual/electric wheelchair with an auxiliary driving wheel, expect-ing to increase the mechanical efficiency and not to enlarge the body of the wheelchair; the module-disassembled drive system is intended for the conven-ience of carrying. Besides, in order to keep both the manual and the electric functions, the caster contains a set of separable orientation control system, while enables the caster to be switched in both the modes of 360-degree rotation and of being controlled by the user. In this way, the use of the wheelchair would be-come much more multiple-choice.

D. Tech. Electrical Engineering. / Developes a wheelchair motion platform whereby its user may be introduced into a simulated world. This simulated world is then required to be closely related to real world spaces that will be encountered by a disabled person using a wheelchair as a mobility aid. The wheelchair to be accommodated in the simulation environment may have multiple mechanical construct possibilities. The wheelchair used on the simulation platform needs to be driven by a combination of two wheels, as is generally found on manual and electric wheelchairs. The final objective was to design the simulation as closely as possible to the real world in order to use the VS-1 motion platform for architectural evaluations, possible training and general research in the field of simulators used in an enabled environment.

M. Tech. Electrical Engineering. / Discusses how to develop the kinematic and dynamic model and the controller for the 2-DOf motion platform used in an augmented reality environment for wheelchair driving. This comes as a motivation to help to train disabled and elderly people to drive wheelchairs.. With accurate inverse dynamic model, it is possible to achieve high performance control algorithms of robots and direct dynamic model is required for their simulation. The other part of this research was to model and control the roller for the feedback of the wheelchair wheels.

M. Tech. Electrical Engineering. / The hypothesis states that a closed loop control system could be applied to a mechatronic system consisting of a mechanical device(s) to be fitted onto the user, and other relevant periphery mechanical devices, appropriately actuated to provide powered movement to assist the wheelchair-bound individual securely attain a standing position.The premise upon which the hypothesis was to be tested relates to the control system's ability to facilitate the upward movement proposed and its degree of stability when simulating the system, thus re-creating the anticipated functionality.

D. Tech. Electrical Engineering. / Proposes a vision-based solution for intention recognition of a person from the motions of the head and the hand This solution is intended to be applied in the context of wheelchair bound individuals whose intentions of interest are the wheelchairs direction and speed variation indicated by a rotation and a vertical motion respectively. Both head-based and hand-based solutions are proposed as an alternative to solutions using joysticks pneumatic switches, etc.

M. Tech. Electrical Engineering / Determines whether Hidden Markov models (HMM) can be used to classify steady state visual evoked electroencephalogram signals in a BCI system. This is for the purpose of aiding disabled people in driving a wheelchair.

M. Tech. Electrical Engineering. / Investigates a new approach of wheelchair control, based on the user behaviour recognition. This objective involves two steps in the resolution of the problem. The first step is to determine the action the user initiates. Therefore, the present study will mostly refer to literatures on car driver behaviour modelling, as several studies have been conducted in that domain. The proposed model of user's behaviour presented here is based on probabilistic graphical model, for instance, Bayesian network. The second step is the generation of an assistive control signal that will compensate the user input, depending on the driving task inferred by the Bayesian network.Experiments have been conducted on a virtual environment model developed in Matlab and several users participated to the experiments. The results show a great potential of Bayesian Network model to infer on human behaviour and also a satisfying output from the ANFIS model as it delivers signals following the user's behaviour.