Optimal Control & LQR/LQG Reading List

Curated by Mouhssine Rifaki | Stanford Electrical Engineering | Last updated August 2026

Linear-quadratic control is the workhorse of modern robotics and the cleanest setting where learning and control meet. From Riccati equations to policy-gradient guarantees, these works shape how we think about stability, robustness, and sample efficiency.

Optimal Control & LQR/LQG: 10 key papers

1. Learning Convex Optimization Control Policies
   Agrawal et al. arXiv 2019.
2. On the Sample Complexity of the Linear Quadratic Regulator
   Dean et al. arXiv 2017.
3. Regret Bounds for Robust Adaptive Control of the Linear Quadratic Regulator
   Dean et al. arXiv 2018.
4. A Tour of Reinforcement Learning: The View from Continuous Control
   Recht. arXiv 2018.
5. Certainty Equivalence is Efficient for Linear Quadratic Control
   Mania et al. arXiv 2019.
6. Naive Exploration is Optimal for Online LQR
   Simchowitz and Foster. arXiv 2020.
7. The Nonstochastic Control Problem
   Hazan et al. arXiv 2019.
8. Logarithmic Regret for Online Control
   Agarwal et al. arXiv 2019.
9. Online Control with Adversarial Disturbances
   Agarwal et al. arXiv 2019.
10. Adaptive Control and Regret Minimization in Linear Quadratic Gaussian (LQG) Setting
    Lale et al. arXiv 2020.

Related reading lists

• Policy Gradient Methods
• Sample Complexity in Reinforcement Learning
• Linear MDPs & Linear Function Approximation
• Adversarial Reinforcement Learning & Robustness to Adversaries
• Robust & Risk-Sensitive RL
• System Identification & Learning Dynamical Systems
• Low-Rank Structure in Machine Learning
• Block MDPs & Latent State Discovery