Core Concepts
The authors introduce Robust CLVFs for nonlinear systems with disturbances, inheriting properties from CLVFs and providing efficient computation techniques.
Abstract
The paper introduces Robust Control Lyapunov-Value Functions (R-CLVF) for systems with bounded disturbances, inheriting properties from CLVFs. It extends the theory to include robust exponential stabilizability and provides numerical examples validating the theory.
Control Lyapunov Functions (CLFs) are widely used in control systems but lack systematic construction methods for general nonlinear systems. The R-CLVF addresses this limitation by defining a method to find a non-smooth CLF for systems with bounded disturbances.
The R-CLVF inherits properties from the CLVF, identifying the smallest robust control invariant set (SRCIS) and stabilizing the system with a user-specified exponential rate. Techniques like warmstart and system decomposition are introduced to address computational challenges.
Numerical examples illustrate the trade-off between decay rate and region of exponential stabilizability, showcasing efficiency using warmstart and decomposition techniques.
Stats
The R-CLVF is Lipschitz continuous, satisfies dynamic programming principle, and is unique viscosity solution to corresponding VI.
The SRCIS is defined as zero-level set of computed R-CLVF.
Different choices of loss function norm result in different SRCIS, ROES, trajectories.