This is the Julia code for Learning-based Warm-Starting for Fast Sequential Convex Programming and Trajectory Optimization, in which a neural network is used to warm-start a trajectory optimization algorithm (GuSTO) leading to faster convergence.

All required packages can be installed by cloning this repository, navigating to the project directory, and using the following command in the Julia REPL package manager:

(v1.0) pkg> activate .

(nnGuSTO) pkg> instantiate

An example notebook can be run through:

jupyter notebook demo_gusto_julia.ipynb 

More detailed analysis can be found in the following notebooks:

jupyter notebook src/notebooks/

• S. Banerjee, T. Lew, R. Bonalli, A. Alfaadhel, I. A. Alomar, H. M. Shageer, and M. Pavone, “Learning-based Warm-Starting for Fast Sequential Convex Programming and Trajectory Optimization,” in IEEE Aerospace Conference, Big Sky, Montana, 2020.