This code is to accompany the paper Imitation Learning with Stability and Safety Guarantees. It learns a Neural Network controller with stability and safety guarantees through imitation learning process.
- He Yin (he_yin at berkeley.edu)
- Peter Seiler (pseiler at umich.edu)
- Ming Jin (jinming at vt.edu)
- Murat Arcak (arcak at berkeley.edu)
The code is written in Python3 and MATLAB.
There are several packages required:
- MOSEK: Commercial semidefinite programming solver
- CVX: MATLAB Software for Convex Programming
- Tensorflow: Open source machine learning platform
To plot the computed ROA, two more packages are required:
- SOSOPT: General SOS optimization utility
- Multipoly: Package used to represent multivariate polynomials
- To start the safe imitation learing process, go to each folder, run NN_policy.py. The computation results are stored in the folder data.
- To visualize the results for the inverted pendulum example, run result_analysis.m. For the GTM and vehicle lateral control examples, run plot_generation.m.
ROAs of the Learned NN Controllers and Explicit MPC Controller for the Vehicle Lateral Control Example
