Implementation of iLQR (Iterative Linear Quadratic Regulator), with example application of swinging up a cart pole from an arbitrary initial state.
See my blog post on iLQR for more details.
To reproduce the results shown in the blog post, execute the following command from the project root:
python ilqr_cartpole_learn.py --horizon 12000 \
--iterations 50 \
--thresh 100 \
--grad_clip 30