Given time series data from an accelerometer and gyroscope attached to the prosthetic lower leg limb of 10 subjects, we create two different models (1D CNN and bidirectional LSTM) to investigate if we can classify the type of terrain a user is traversing.