Stroke is the second largest cause of disability in the world. Disability in the upper limb in particular, affects patient independence. Both mental and physical rehabilitation are important for effective recovery after a stroke. Hybrid exoskeletons are a recent development which combine Functional Electrical Stimulation (FES) with actuators to improve both the mental and physical rehabilitation of stroke patients. FES is the application of electric pulses to the muscle which cause the muscle to contract. FES is useful in stroke rehabilitation as it causes the patient’s muscles to perform work and thus helps to build muscle strength. However when used on its own FES is difficult to control and causes rapid muscle fatigue. Assistive exoskeletons using various actuators, most commonly electric motors, have been used in stroke rehabilitation to help patient’s perform precise movements which aid neural repair. However these exoskeletons tend to be bulky and heavy, and they do not improve the muscle strength of the patient on their own.

Hybrid exoskeletons (exoskeletons which combine FES and an actuator) have been shown capable of reducing the weight of the actuator and improving movement precision compared to FES alone. However, little attention has been given towards optimising the control between the actuator and the FES with regards to reduction and management of FES-induced fatigue or with regards to adapting to user ability. FES control systems used for upper limb hybrid exoskeletons simply ramp up stimulation intensity when fatigue is observed. FES control systems used outside of hybrid exoskeletons tend towards the other extreme where complex models requiring many measurements for model parameter estimates result in setup times which are unsuitable for clinical applications. Recent research suggests that controlling multiple FES input parameters (amplitude, pulse-width, and frequency) may reduce FES-induced fatigue however no model exists which relates all three of these FES input parameters to muscle movement. This thesis introduces a novel model which relates all three FES input parameters to rotation of the elbow in the sagittal plane through stimulation of the bicep muscle for use in an upper limb assist-as-needed hybrid exoskeleton.

During this work, advice was sought from local stroke rehabilitation professionals and observations of stroke rehabilitation sessions were undertaken. The first chapter in this work highlights the needs as emphasized by stroke rehabilitation professionals and as observed by the author during these observation sessions. These needs and requirements are kept in mind throughout the entirety of this work. Following the background chapter, a thorough literature review is presented, which gives a detailed overview of the current upper limb hybrid exoskeletons. Gaps in the field are highlighted and the potential advantages of hybrid exoskeletons are described.

This work includes the construction of a voltage-controlled FES device which allows control of multiple FES parameters. The device is tested using two different types of electrodes, hydrogel electrodes, and washable e-textile electrodes. The e-textile electrodes were found to be more consistent over multiple uses and capable of inducing movement in the Bicep muscle at a lower voltage than the hydrogel electrodes due to a lower resistance. Thus the e-textile electrodes have been used throughout the rest of the work. Following the testing of the FES device and electrodes, the novel FES model is introduced and tested on 8 healthy subjects using several combinations of the FES input parameters. Results show that the model is able to predict the change in angle in response to FES inputs with an R² value of 0.66 (no auto-regressive behaviour) across all subjects and all parameter combinations. Predictions of the response to the combination FES parameter inputs and to pulse-width parameter steps were the best (R² = 0.63 –> 0.77), while predictions for voltage steps and frequency steps were the worst (R² = 0.37 and 0.28 respectively). Only two measurements were required (the voltage threshold and the overall gain) to be obtained for a specific subject.

An electric motor was selected as the actuator due to portability and a good power to weight ratio. The placement of the actuator was on the shoulder to avoid placing too much weight on the arm, however in future it may be better to place the actuator on the back of the user as the weight of the exoskeleton is still likely too heavy for extended use on a stroke patient given common issues with subluxation in the shoulder of many stroke patients. A load sensor and an angle sensor are used to measure the movement and interaction of forces between the exoskeleton and the user.

The feedback from the sensors and the novel multi-parameter-controlled FES model are then implemented in an assist-as-needed control scheme. Testing of the model once implemented in the overall assist-as-needed control system showed that the overall gain for the FES model could be measured during runtime further reducing the setup time of the device. The entire setup of the hybrid exoskeleton took only a few minutes and could be performed by a healthy subject using only one arm. Testing of the entire hybrid exoskeleton and assist-as-needed control platform was performed on one healthy subject and found to produce similar tracking errors to that of the motor only control. The hybrid system produced 24˚ less average angle error and 13.21˚ less RMSE, than FES on its own and showed a reduction in FES-induced fatigue. Hybrid exoskeletons offer many advantages in the area of stroke rehabilitation and more research should be conducted in order to fully realise their benefits.

This work presents the first upper limb assist-as-needed hybrid exoskeleton. It consists of a novel multi-parameter-controlled FES model. The FES model has been tested on several healthy subjects, and implemented in the hybrid exoskeleton assist-as-needed control system which was then tested on one healthy subject. The model and method of control of FES described in this work has been specifically designed to take advantage of the benefits of hybrid exoskeletons. It allows for more control of the FES response as compared to other hybrid exoskeletons while reducing the FES-induced fatigue and complexity compared with non-hybrid, FES-only systems. This work provides a platform on which further testing could be conducted towards better understanding and reducing FES-induced fatigue and which could be extended and built upon to further improve stroke rehabilitation patient gains, as well as improve monitoring of patient ability, and reduce the hand-on time of therapists. This is the first research which has investigated (and demonstrated) the ability of hybrid exoskeletons to reduce FES-induced fatigue.