Learning to Control Known Feedback Linearizable Systems From Demonstrations

• 成果类型: Article
• 署名作者: Sultangazin, Alimzhan; Pannocchi, Luigi; Fraile, Lucas; Tabuada, Paulo
• 署名单位: University of California System; University of California Los Angeles; University of California System; University of California Los Angeles; Scuola Superiore Sant'Anna
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2023.3272392
• 发表日期: 2024
• 页码: 189-201
• 关键词: trajectory; control systems; Task analysis; cloning; asymptotic stability; Transforms; STANDARDS; Machine Learning; Motion control; nonlinear control systems; robot control

摘要

this article, we revisit the problem of learning a stabilizing controller from a finite number of demonstrations by an expert. By first focusing on feedback linearizable systems, we show how to combine expert demonstrations into a stabilizing controller, provided that demonstration trajectories are sufficiently long and there are at least n + 1 of them, where n is the number of states of the system being controlled. When we have more than n + 1 demonstration trajectories, we discuss how to optimally choose the best n + 1 demonstrations to construct the stabilizing controller. We then extend these results to a class of systems that can be embedded into a higher dimensional system containing a chain of integrators. The feasibility of the proposed algorithm is demonstrated by applying it on a CrazyFlie 2.0 quadrotor.