Enforcing robust control guarantees within neural network policies
This work addresses the problem of ensuring safety and performance in control systems for practitioners, representing an incremental advance by integrating existing techniques.
The paper tackles the tradeoff between robustness and performance in safety-critical control systems by proposing a method that combines neural network policies with robust control guarantees, achieving improved average-case performance over robust control methods and better worst-case stability compared to non-robust deep RL methods.
When designing controllers for safety-critical systems, practitioners often face a challenging tradeoff between robustness and performance. While robust control methods provide rigorous guarantees on system stability under certain worst-case disturbances, they often yield simple controllers that perform poorly in the average (non-worst) case. In contrast, nonlinear control methods trained using deep learning have achieved state-of-the-art performance on many control tasks, but often lack robustness guarantees. In this paper, we propose a technique that combines the strengths of these two approaches: constructing a generic nonlinear control policy class, parameterized by neural networks, that nonetheless enforces the same provable robustness criteria as robust control. Specifically, our approach entails integrating custom convex-optimization-based projection layers into a neural network-based policy. We demonstrate the power of this approach on several domains, improving in average-case performance over existing robust control methods and in worst-case stability over (non-robust) deep RL methods.