CDC 2026 Learning-Based Control Tutorial

Tutorial session · CDC 2026 · Honolulu, HI, USA

A Decade of Learning-Based Control: From Theoretic Foundations to Practical Deployment

When
Tuesday, December 15, 2026
10:00 – 11:45
Where
Coral 3-5 · Session TuAAT21
Hilton Hawaiian Village, Honolulu
Paper
Coming soon

Overview

Building on ten years of Learning-Based Control Invited Sessions at the IEEE Conference on Decision and Control (CDC), this tutorial reviews key advances at the intersection of machine learning and control. It focuses on methods that combine the flexibility of machine learning with the structure and theoretical foundations of control, with the goal of designing autonomous systems that are capable, reliable, and safe. The tutorial opens with core notions for establishing stability and safety guarantees, then reviews approaches that learn dynamics models to improve control under uncertainty, and turns to data-driven methods that synthesize or learn controllers directly from data, including direct data-driven control, reinforcement learning, imitation learning, and Bayesian optimization. It closes with a practical example and a discussion of open challenges. Throughout, the aim is to offer an accessible entry point to the field and to support continued progress toward reliable learning-based control in real-world systems. A companion tutorial paper, along with tutorial resources and codebases, is available for further reading and exploration (see Resources).

Prior model
Data
  1. Dynamics Learning for Control
  2. Direct Data-Driven Control
  3. Reinforcement Learning, Imitation Learning, and Bayesian Optimization
Closed-Loop Stability and Safety Guarantees
Applications
  • Robotics
  • Healthcare
  • Energy and power systems
  • Process control
  • …
Scope of the tutorial. Learning-based control combines prior models and data with control-theoretic structure. The tutorial first introduces foundations for closed-loop stability and safety guarantees, then covers methods ranging from structured dynamics learning to increasingly data-driven policy learning, and closes with representative applications.

Schedule

All times are local Honolulu time (Hawaii–Aleutian Standard Time, UTC−10).

Time Session
10:00 – 10:10 Introduction: A Decade of Learning-Based ControlAngela Schoellig, Sebastian Trimpe, Melanie Zeilinger, and Matthias Müller
10:10 – 10:30 Stability and SafetyMelanie Zeilinger
10:30 – 10:50 Dynamics Learning for ControlAngela Schoellig and SiQi Zhou
10:50 – 11:10 Direct Data-Driven ControlMatthias Müller
11:10 – 11:30 Reinforcement Learning, Imitation Learning, and Bayesian OptimizationSebastian Trimpe
11:30 – 11:45 Practical Example and Future DirectionsAngela Schoellig

Organizers and Contributors

Resources

Tutorial Paper, Slides, and Recordings

Codebases