Corpus ID: 220041530

Adaptive Control and Regret Minimization in Linear Quadratic Gaussian (LQG) Setting

  title={Adaptive Control and Regret Minimization in Linear Quadratic Gaussian (LQG) Setting},
  author={Sahin Lale and K. Azizzadenesheli and B. Hassibi and Anima Anandkumar},
  journal={arXiv: Learning},
  • Sahin Lale, K. Azizzadenesheli, +1 author Anima Anandkumar
  • Published 2020
  • Computer Science, Mathematics
  • arXiv: Learning
  • We study the problem of adaptive control in partially observable linear quadratic Gaussian control systems, where the model dynamics are unknown a priori. We propose LqgOpt, a novel reinforcement learning algorithm based on the principle of optimism in the face of uncertainty, to effectively minimize the overall control cost. We employ the predictor state evolution representation of the system dynamics and deploy a recently proposed closed-loop system identification method, estimation, and… CONTINUE READING
    4 Citations
    Explore More and Improve Regret in Linear Quadratic Regulators
    • 3
    • PDF


    Regret Bounds for the Adaptive Control of Linear Quadratic Systems
    • 184
    • Highly Influential
    • PDF
    Thompson Sampling for Linear-Quadratic Control Problems
    • 37
    • PDF
    Model-Free Linear Quadratic Control via Reduction to Expert Prediction
    • 58
    • PDF
    Certainty Equivalence is Efficient for Linear Quadratic Control
    • 43
    • PDF
    Certainty Equivalent Control of LQR is Efficient
    • 49
    • PDF