Neural Lyapunov Redesign
This addresses safety concerns in control systems for physical and non-physical tasks, but it is incremental as it builds on existing Lyapunov-based methods.
The paper tackles the problem of ensuring stability as a safety guarantee in learning-based controllers by combining Lyapunov function estimation with automatic controller synthesis to enlarge safe regions. It provides theoretical results on applicable systems and empirically evaluates the method on an exemplary dynamical system.
Learning controllers merely based on a performance metric has been proven effective in many physical and non-physical tasks in both control theory and reinforcement learning. However, in practice, the controller must guarantee some notion of safety to ensure that it does not harm either the agent or the environment. Stability is a crucial notion of safety, whose violation can certainly cause unsafe behaviors. Lyapunov functions are effective tools to assess stability in nonlinear dynamical systems. In this paper, we combine an improving Lyapunov function with automatic controller synthesis in an iterative fashion to obtain control policies with large safe regions. We propose a two-player collaborative algorithm that alternates between estimating a Lyapunov function and deriving a controller that gradually enlarges the stability region of the closed-loop system. We provide theoretical results on the class of systems that can be treated with the proposed algorithm and empirically evaluate the effectiveness of our method using an exemplary dynamical system.