Efficient Online Linear Control with Stochastic Convex Costs and Unknown Dynamics
This work addresses the challenge of efficient online control for unknown linear systems, which is important for applications like robotics and autonomous systems, though it appears incremental as it builds on existing regret frameworks with a new algorithmic approach.
The paper tackles the problem of controlling an unknown linear dynamical system with stochastic convex costs and full feedback, presenting an algorithm that achieves optimal √T regret compared to the best stabilizing linear controller in hindsight. The result is based on the Optimism in the Face of Uncertainty paradigm, leading to improved computational efficiency and simpler analysis.
We consider the problem of controlling an unknown linear dynamical system under a stochastic convex cost and full feedback of both the state and cost function. We present a computationally efficient algorithm that attains an optimal $\sqrt{T}$ regret-rate compared to the best stabilizing linear controller in hindsight. In contrast to previous work, our algorithm is based on the Optimism in the Face of Uncertainty paradigm. This results in a substantially improved computational complexity and a simpler analysis.