Fig. 1
From: Generating synthetic gait patterns based on benchmark datasets for controlling prosthetic legs
Overview of the proposed method. First, a GAN is trained using benchmark datasets. Second, a synthetic gait pattern is generated from user input using the GAN. Then, the set of gait patterns is converted into time-series data; the corresponding data for mechanical impedance parameters are generated. Finally, a DNN model that generates impedance parameters for controlling a prosthetic leg is trained on the generated time-series data