The paper focuses on the analysis and design of safety-critical control for stochastic systems using control barrier functions (CBFs). It makes the following key contributions:
Proposes an almost sure reciprocal control barrier function (AS-RCBF) and an almost sure zeroing control barrier function (AS-ZCBF) that ensure the safety of a set with probability one for stochastic systems. The conditions for these CBFs are more relaxed compared to previous work.
Introduces a new type of stochastic zeroing control barrier function (Stochastic ZCBF) that directly incorporates the diffusion coefficients of the stochastic system. This allows evaluating the probability of a trajectory staying in a safe set.
Provides control design strategies using the proposed AS-RCBF, AS-ZCBF, and Stochastic ZCBF. The control laws are shown to ensure safety with probability one or a bounded probability.
Demonstrates the validity of the proposed CBFs and control designs through simple examples with numerical simulations.
The paper highlights the challenges in ensuring almost sure safety in stochastic systems and proposes new CBF-based approaches to address this. The results provide a framework for designing safety-critical controllers for a wide range of stochastic systems.
לשפה אחרת
מתוכן המקור
arxiv.org
תובנות מפתח מזוקקות מ:
by Yuki Nishimu... ב- arxiv.org 04-18-2024
https://arxiv.org/pdf/2209.08728.pdfשאלות מעמיקות