Development of a novel hand-held haptic device integrating upper limb movement assessment and directional guidance

Visual impairment is one of the most common symptoms after several diseases including cataracts, diabetes, and stroke. It can cause severe impact on the quality of life such as a decrease in workforce participation and productivity and an increase in the chance of depression among all ages. This effect is even worse on patients after stroke because it prevents the use of advanced robotic devices. Most of the current upper limb robotic devices rely on visual cues to guide the movement throughout the rehabilitation process. Therefore, it is beneficial to design a local navigation device based on haptic cues for visually impaired patients after stroke. With the development of sensing techniques, it is also possible to integrate movement assessment function based on kinematic sensor data, which can be more objective, sensitive, and continuous than traditional assessments.
This research aims to design and develop a novel hand-held haptic device for movement guidance and movement assessment for robot-assisted rehabilitation outside clinical environments, especially for visually impaired people after stroke.
The movement assessment was conducted on kinematic data collected from a position sensor or an accelerometer. Two novel position sensing methods were proposed, and several kinematic features were extracted from the measurements to objectively quantify movement smoothness with the help of machine learning. An observational experiment was finally conducted to verify the effectiveness of kinematic assessment. The results showed that kinematic features could reflect subtle progress in motor function learning progress and could contribute to the machine learning models development for a better classification result on both movement type and movement smoothness.
The design of the haptic implementation was firstly explored with three different haptic motors, among which a voice coil actuator was selected to generate asymmetric vibrations for haptic delivery. The input control signal was then parameterised as the main contribution, and five output parameters were discussed. A psychophysical experiment was finally conducted to find the ideal characteristics of input signal that could produce clearer haptic directional cues. The results showed that input signals after optimisation could improve the delivery of haptic directional cues in terms of accuracy, applicability, and user’s confidence.