meta_learn_mpc_opt

This repo contains experiments to speed up differentiable MPC optimization by learning a meta-learner that learns an optimizer which optimizes the cost function with few iterations of backpropagation.

Meta-Learners tested so far

• MAML
• Meta-LSTM
• Iterative Optimizer

Control tasks

• Double Integrator

Credits

• https://github.com/markdtw/meta-learning-lstm-pytorch
• https://github.com/learnables/learn2learn
• https://github.com/facebookresearch/higher