GAN-MPC: Training Model Predictive Controllers with Parameterized Cost Functions using Demonstrations from Non-identical Experts

Published in "RSS 2023 Workshop"
Returaj Burnwal , Anirban Santara , Nirav P.Bhatt , Balaraman Ravindran , Gaurav Aggarwal

Model Predictive Control (MPC) is widely used for trajectory optimization in robotics. This paper introduces a novel approach, using a generative adversarial network (GAN), to train Learnable-MPC policy when the demonstrator and imitator have different dynamics, achieving effective behavior learning without direct action copying.