AStanford AA203 Optimal and Learning-Based Control | Spring 2026 | Lecture 8: Nonlinearity

This lecture provides a comprehensive and rigorous introduction to using LQR for nonlinear control, a topic of central importance in robotics and autonomous systems. Prof. Pavone’s expertise is evident in the clarity of the derivations and the practical insights he shares, such as the two-step design philosophy and the importance of regularization in ILQR. The content is well-structured, building from a recap of LQR to its extensions for tracking and trajectory optimization. The mathematical derivations are thorough, and the lecture effectively bridges theory and practice by discussing implementation details and common pitfalls. The use of a companion textbook and course materials enhances the educational value. However, the lecture is primarily a teaching resource and does not present novel research findings. It also assumes a solid background in optimal control and linear algebra, which may limit accessibility to a broader audience. The discussion of ILQR and DDP is particularly valuable, as these are widely used algorithms in practice. The comparison with SCP provides useful context for choosing appropriate methods. Overall, this is an excellent lecture that achieves its educational objectives with high rigor and clarity.

The title accurately reflects the content: Lecture 8 focuses on nonlinearity in optimal control, covering tracking LQR, iterative LQR, and DDP.

9/10

Lecture by a renowned expert (Prof. Marco Pavone) from Stanford University, part of a formal course. Content is rigorous, well-structured, and based on established optimal control theory. The lecture includes derivations, references to a textbook, and practical insights. Minor limitations: no explicit citations to external sources beyond the course materials, and the lecture is an educational presentation rather than a peer-reviewed publication.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 00:00Introduction and recap of dynamic programming and LQR from previous lecture.
• 05:00Discussion on the generality of LQR and its role in control hierarchy.
• 10:00Recap of LQR formulation and generalization with cross terms and affine dynamics.
• 20:00Introduction to trajectory tracking using LQR for linear systems.
• 30:00Extension to nonlinear systems via linearization and formulation of auxiliary LQR problem.
• 40:00Introduction to iterative LQR (ILQR) for trajectory optimization.
• 50:00Detailed explanation of ILQR backward and forward passes, and comparison with SCP.
• 1:00:00Discussion of practical considerations for ILQR: regularization, line search, initialization.
• 1:05:00Introduction to Differential Dynamic Programming (DDP) and its relation to ILQR.
• 1:10:00Summary and conclusion of the lecture.

• AA203 Optimal and Learning-Based Control course page — Course information and enrollment details.
• Principles of Robot Autonomy (companion textbook) — Free online textbook referenced as companion reading.
• AA203 Spring 2026 course schedule and syllabus — Course schedule and syllabus.
• Lecture 8 slides — Slides used in the lecture.
• AA203 full playlist — Playlist of all lectures in the course.

• Principles of Robot Autonomy — Companion textbook that likely covers similar material in more depth.