Exploring the Potential of Passive Upper Limb Rehabilitation through the Use of Inertial Sensors and Haptic Technology

Abstract

It is irrefutable that the role of technology in the area of rehabilitation is increasing rapidly. The ability to provide quantification in an area previously governed by subjective measures is forecasted to revolutionize the healthcare practice. The technology also promises to increase autonomy with therapy routines and create more immersive and enjoyable sessions through the integration of gaming interfaces. This research focused on the development and validation of a wearable device capable of measuring movement of the upper limb and providing passive feedback through vibrotactile technology. The upper limb movement was captured through inertial measurement unit (IMU) sensors, whose output were used to construct a real-time kinematic model of the upper limb. Outputs from this kinematic model were used to derive vibrotactile feedback through a proportional controller. The system’s ability to measure movement of the upper limb was quantified by an experiment which compared the output of the system to a camera based motion capture system utilizing an approved upper limb kinematic model. Participants in the experiment conducted 15 different activities of daily living (ADL’s) to assess the system in a rehabilitative context. Results from this experiment showed a high correlation between the two systems outputs. Absolute accuracy was seen to be movement dependent and ranged from overall root mean squared (RMS) error of 6 to 20 degrees. To validate the concept of vibrotactile cues as a form of movement feedback, an experiment was conducted which compared the ability of vibrotactile cues to a visual interface as a source of instruction for carrying out a simple task. Metrics that are usually reserved for measuring the performance of control systems such as rise time, settling time, overshoot percentage and delay time were derived from the data to quantify performance. Results showed, that when tuned correctly, vibrotactile actuation could compare to visual feedback in terms of overall user performance. It also highlighted that the control strategy used to generate the vibrotactile cue had a significant effect on user performance. Finally, based on the verification that vibrotactile cues do provide a viable feedback mechanism, it un-covered the need for a greater level of understanding of the human-vibrotactile interaction. Various models were constructed to represent this interaction with the final being optimized based on experimental data obtained. The resulting simulation model was then placed in a closed loop situation with a controller similar to that implemented experimentally. Results of how the model performed under different control conditions are discussed with future suggestions in regards to those which should be implemented experimentally. The system designed and developed by this masters has laid the foundations for a new research area and provides a platform for further development in the field.