Safe and Near-Optimal Policy Learning for Model Predictive Control using Primal-Dual Neural Networks

Primal-Dual Policy Learning Simple Example approxMPCTrain.m file trains the Primal and Dual policies. This must be run to generate and store all data. approxMPCTest.m file then loads the trained policies and data and runs test on new samples. If test statistics are deemed unsatisfactory, please go back to training code and retrain

Requirements

Multiparametric toolbox with explicit MPC solver. YALMIP. Gurobi. MATLAB 2016b or newer with Neural Network Toolbox.

Theory

Safe and Near-Optimal Policy Learning for Model Predictive Control using Primal-Dual Neural Networks, [https://arxiv.org/abs/1906.08257]

Contacts

Monimoy Bujarbaruah (monimoyb@berkeley.edu) and Xiaojing Zhang (xiaojing.zhang@berkeley.edu)