ANFIS based knee angle prediction: An approach to design speed adaptive contra lateral controlled AK prosthesis

Prediction of prosthetic knee joint angle plays a crucial role in performance of above knee (AK) prosthesis, as it helps in estimating intended posture and movement of amputee. This paper applies a first-order Sugeno type ANFIS (Adaptive Neuro-Fuzzy Inference System) to predict the knee angle from contra lateral knee angle with different walking speed. The other two variables to train ANFIS were derivative of knee angle trajectory and the best fit curve equation between the trajectories of both knee angles. The average RMSE was 3.4 ± 1.4° with wide range of walking speeds, using few Fuzzy rules. This research work has direct application in the design of a contra lateral controlled speed adaptive AK prosthesis.