System Identification Based on Generalized Orthonormal Basis Function for Unmanned Helicopters: A Reinforcement Learning Approach

System identification is very important for the controller design of unmanned helicopters, and it has significant impacts on the quality of flight missions. With the recent advances of reinforcement learning, we propose a generalized orthonormal basis function (GOBF)-based system identification scheme for unmanned helicopters. Using GOBF, the traditional parameter estimation problem in system identification not only becomes better numerical conditioned but also can be affected by the prior knowledge. The proposed novel GOBF-based system identification scheme can enable users to make good use of prior knowledge due to the learning capability of reinforcement learning. In addition, there is a theoretical guarantee of convergence in the proposed GOBF-based system identification scheme. Moreover, we develop an additional self-optimization process based on Bayes theory to enhance the robustness of the scheme and improve the search efficiency of the proposed scheme. Both simulations and practical experiments show the effectiveness and advantages of our proposed scheme.