This week progress continues mainly on two fronts: calculation of FOM for both the FollowerStopper and PI with saturation, and implementation of a deep learning model for computation of the characteristic velocity. For the FOM, we are doing more simulations of the FollowerStopper when the command velocity, \(U\), is in some small neighborhood of the characteristic velocity. Additionally we are doing simulations of the PI controller to generate plots similar to those shown in last weeks blog post.

For the deep learning, I working with our newest member, Ray Zhang. Once we have collected more simulation data, Ray Zhang will look into using LSTM's for identifying the characteristic velocity while I will be implementing a feedforward neural net.