EE364b Final Project
Tarun Punnoose and Nathan Kau
This is a ADMM-based trajectory optimization solver for problems with time-varying linear dynamics, quadratic state costs, control constraints, and group lasso (L2) cost on the control. The L2 control penalty encourages bang-off-bang control.
Please see "Paper.pdf" located in this repo for more detail.
To run the ADMM-based solver on an example satellite rendezvous problem:
include("src/Main.jl")To run the CVX solver as a comparison (which doesn't converge):
include("src/CVXTest.jl")