Safe Control Under Input Limits with Neural Control Barrier Functions
Safe Control Under Input Limits with Neural Control Barrier Functions
复制标题
DOI：
10.48550/arxiv.2211.11056
复制
发表时间：
2022-11
影响因子：
4.6
通讯作者：
Simin Liu;Changliu Liu;J. Dolan
中科院分区：
文献类型：
--
作者：
Simin Liu;Changliu Liu;J. Dolan
我们提出了新的方法来合成控制障碍函数（CBF）为基础的安全控制器，避免输入饱和，这可能会导致安全违规。特别是，我们的方法是为高维，一般的非线性系统，这样的工具是稀缺的。我们利用机器学习技术，如神经网络和深度学习，来简化非线性控制设计中的这一挑战性问题。该方法由一个学习者-评论家架构，其中的评论家给出输入饱和的反例和学习者优化神经CBF，以消除这些反例。我们提供了经验的10维状态，四维输入四轴飞行器摆系统的结果。我们学习的CBF避免了输入饱和，并在近100%的试验中保持安全性。
We propose new methods to synthesize control barrier function (CBF)-based safe controllers that avoid input saturation, which can cause safety violations. In particular, our method is created for high-dimensional, general nonlinear systems, for which such tools are scarce. We leverage techniques from machine learning, like neural networks and deep learning, to simplify this challenging problem in nonlinear control design. The method consists of a learner-critic architecture, in which the critic gives counterexamples of input saturation and the learner optimizes a neural CBF to eliminate those counterexamples. We provide empirical results on a 10D state, 4D input quadcopter-pendulum system. Our learned CBF avoids input saturation and maintains safety over nearly 100% of trials.