Using Artificial Intelligence To Improve The Control Of Prosthetic Legs
For as long as people have been able to survive limb threatening injuries prostheses have been created. Modern lower limb prostheses are primarily controlled by adjusting the amount of damping in the knee to bend in a suitable manner for walking and running. Often the choice of walking state or running state has to be controlled manually by pressing a button. While this simple tuning strategy can work for many users it can be limiting and there is the tendency that controlling the leg is not intuitive and the wearer has to learn how to use leg. This thesis examines how this control can be improved using Artificial Intelligence (AI) to allow the system to be tuned for each individual. A wearable gait lab was developed consisting of a number of sensors attached to the limbs of eight volunteers. The signals from the sensors were analysed and features were extracted from them which were then passed through 2 separate Artificial Neural Networks (ANN). One network attempted to classify whether the wearer was standing still, walking or running. The other network attempted to estimate the wearer’s movement speed. A Genetic Algorithm (GA) was used to tune the ANNs parameters for each individual. The results showed that each individual needed different parameters to tune the features presented to the ANN. It was also found that different features were needed for each of the two problems presented to the ANN. Two new features are presented which identify the movement states of standing, walking and running and the movement speed of the volunteer. The results suggest that the control of the prosthetic limb can be improved.
- PhD