Poster
Complexity of Derivative-Free Policy Optimization for Structured Control
Xingang Guo · Darioush Keivan · Geir Dullerud · Peter Seiler · Bin Hu
Great Hall & Hall B1+B2 (level 1) #1208
Abstract:
The applications of direct policy search in reinforcement learning and continuous control have received increasing attention.In this work, we present novel theoretical results on the complexity of derivative-free policy optimization on an important class of robust control tasks, namely the structured synthesis with static output feedback. Optimal synthesis under structural constraints leads to a constrained nonconvex nonsmooth problem and is typicallyaddressed using subgradient-based policy search techniques that are built upon the concept of Goldstein subdifferential or other notions of enlarged subdifferential. In this paper, we study the complexity of finding -stationary points for such nonsmooth robust control design tasks using policy optimization methods which can only access the zeroth-order oracle (i.e. the norm of the closed-loop system). First, we study the exact oracle setting and identify the coerciveness of the cost function to prove high-probability feasibility/complexity bounds for derivative-free policy optimization on this problem. Next, we derive a sample complexity result for the multi-input multi-output (MIMO) -norm estimation. We combine this with our analysis to obtain the first sample complexity of model-free, trajectory-based, zeroth-order policy optimization on finding -stationary points for structured control. Numerical results are also provided to demonstrate our theory.
Chat is not available.