Lecture 23: MIT 6.832 Underactuated Robotics (Spring 2022) | "Output Feedback"

Lecture 23: MIT 6.832 Underactuated Robotics (Spring 2022) | "Output Feedback"

by underactuated

Formal & Physical Sciences Control & Robotics TJFElectronics engineeringTJFMAutomatic control engineering
🎙 underactuated 👥 17K 📅 May 4, 2022 ⏱ 76 min 👁 2K 📄 lecture 🧭 2026-08-05
Available in: English (current) Français

Keywords

output feedbackstatic output feedbackobserverseparation principleLQR

Summary

This lecture from MIT’s Underactuated Robotics course focuses on output feedback control, where the controller only has access to measurements y, not the full state x. The professor begins by contrasting static output feedback (u = π(y)) with state feedback, illustrating with a simple linear example that the set of stabilizing gains can be disconnected, making optimization difficult. He cites a result by Blondel and Tsitsiklis showing that static output feedback stabilization is NP-hard in general. The lecture then introduces observer-based feedback, presenting the Luenberger observer as a dynamical system that estimates the state from measurements. The separation principle is discussed, showing that for linear systems, the observer and controller can be designed independently. The professor emphasizes that controllers themselves are dynamical systems with internal state, and he hints at extensions to nonlinear systems and the challenges of output feedback in that context. The lecture concludes with a discussion of the practical implications and the importance of understanding the limitations of output feedback.

163 words

Critical Evaluation

The lecture provides a rigorous and insightful introduction to output feedback control, a topic often overlooked in introductory control courses. The professor’s use of a simple linear example to illustrate the disconnectedness of stabilizing gains is effective and highlights the fundamental difficulty of static output feedback. The reference to the NP-hardness result by Blondel and Tsitsiklis is appropriate and adds credibility, though the professor wisely cautions against overgeneralizing this result to all practical scenarios. The transition to observer-based feedback is natural, and the explanation of the Luenberger observer is clear, with the separation principle correctly presented for linear systems. The lecture’s strength lies in its pedagogical approach: it builds intuition through examples and connects theoretical results to practical implications. However, the lecture could benefit from more explicit references to specific papers or textbooks, as the video description lacks citations. The discussion of nonlinear output feedback is brief, and the professor only hints at the challenges, which might leave advanced students wanting more. Overall, the lecture is of high quality, suitable for graduate-level students, and provides a solid foundation for further study in output feedback control.

185 words

Title / Content Match

The title accurately reflects the content: a lecture on output feedback in underactuated robotics, covering static output feedback, observers, and separation principle.

Quality & Reliability

8/10

Lecture from MIT OpenCourseWare, presented by a professor (likely Russ Tedrake) with rigorous mathematical derivations and references to known results (e.g., Blondel & Tsitsiklis NP-hardness). The content is well-structured and pedagogically sound, though it lacks explicit citations in the video itself.

Key Moments

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

Cited Sources

  • Blondel, V., & Tsitsiklis, J. N. (1997). A survey of computational complexity results in systems and control. — Referenced for NP-hardness of static output feedback.

Concurring Sources

  • Blondel, V., & Tsitsiklis, J. N. (1997). A survey of computational complexity results in systems and control. — Supports the NP-hardness claim for static output feedback.

Contribution & Novelties

The lecture provides a clear and accessible explanation of output feedback control, emphasizing the conceptual shift from static state feedback to dynamic output feedback. It highlights the NP-hardness of static output feedback, a result often not covered in introductory courses, and introduces the observer-based approach as a practical alternative. The lecture also stresses the importance of viewing controllers as dynamical systems, which is a valuable perspective for understanding more advanced topics.

Pour aller plus loin :

  • Separation principle — Relevant for understanding the independence of observer and controller design.
  • Luenberger observer — Directly related to the observer-based feedback discussed.
  • NP-hardness in control — For further reading on complexity results in control.

111 words

Radar Profile

The radar profile shows high scores in technical level and information quality, reflecting the advanced and rigorous nature of the lecture. The quantity of information is also high, but the fiabilite_globale is slightly lower due to the lack of explicit citations in the video itself.

Reliability 8/10