At the 2022 International Conference on Robotics and Automation (ICRA), we are presenting a machine learning model that predicts stance phase and stance/swing transitions for Dephy’s exoboots.  This model works in both unpowered and powered conditions, as long as the powered conditions follow reasonably closely to the torque vs. stance phase presented in the paper. It was trained on multiple speeds, and has been validated with new users (users that the model was not trained on) walking at speeds ranging from 0.7 to 1.7 m/s. 

For the model to work correctly, please follow instructions in the readme on the GitHub page to correctly rotate IMUs and zero ankle angle.