In this report a 1 DOF upper limb rehabilitation exoskeleton was developed. The mechanical design of the exoskeleton was modified to comply with design requirements. The skeletal structure of the exoskeleton consisted of four aluminium links, four hook and loop fasteners and a single potentiometer fixed to the joint of the exoskeleton through its rotational axis. Two pneumatic McKibben artificial muscles were designed and developed for the purpose of actuating the exoskeleton. To fabricate the most suitable McKibben artificial muscles, 3 different experiments were devised to analyse the behaviour of the actuators with respect to; effects of hysteresis, number of turns (n) of the outer shell’s nylon threads and the thickness of the inner bladder used. After fabrication of the actuators, the mathematical model of the contraction of the actuators based on the exoskeleton’s elbow joint angle was derived. EMG signals were used as the primary input for the control scheme. Initially the EMG signals were amplified three times with gains (G) of 10, -15, and -1 the signal was then filtered through an active high pass and a low pass filter with cut off frequencies (fc) of 106.1Hz and 1.98Hz. MATLAB Simulink was used to further process the filtered signal using RMS block and Butterworth low pass filter. To control the exoskeleton a pneumatic actuation system was set up and 3 different control strategies of; single threshold, PID and proportional EMG control methods were implemented. Using single threshold and PID control, did not allow the user to control the range of motion of the exoskeleton. However, the proportional EMG Control allowed the user to control the exoskeleton using EMG signals. In this strategy the EMG signal from the biceps of a specific user (21 year old male) was mapped against corresponding elbow angles, and the data was used in the controller. The exoskeleton was also capable of amplifying user’s power output by a factor of 2 since the EMG measurements while wearing the exoskeleton (240) was approximately halved (119) compared to when the exoskeleton wasn’t worn. The results obtained showed that with correct calibration. The exoskeleton could be controlled with proportion to any user’s EMG signals.
Actuation system
The image below shows a summary of the actuation system: