Title:Data-driven control and transfer learning using neural canonical control structures*

Abstract:An indirect data-driven control and transfer learning approach based on a data-driven feedback linearization with neural canonical control structures is proposed. An artificial neural network auto-encoder structure trained on recorded sensor data is used to approximate state and input transformations for the identification of the sampled-data system in Brunovsky canonical form. The identified transformations, together with a designed trajectory controller, can be transferred to a system with varied parameters, where the neural network weights are adapted using newly collected recordings. The proposed approach is demonstrated using an academic and an industrially motivated example.