Lecture 4 | MIT 6.832 (Underactuated Robotics), Spring 2020 | Continuous Dynamic Programming

The lecture provides a rigorous and clear exposition of continuous dynamic programming, building on discrete concepts. The instructor’s pedagogical approach is effective, using the double integrator example to illustrate key ideas and limitations. The derivation of the HJB equation is well-motivated, and the discussion of its sufficiency but not necessity is scientifically accurate. The lecture acknowledges the challenges of continuous methods, such as the need for differentiability and the difficulty of representing arbitrary value functions, which is honest and valuable for students. The content is highly technical and assumes prior knowledge of control theory and optimization, but it is appropriate for an advanced undergraduate or graduate course. The sources cited are limited to the course website, which is a minor weakness, but the lecture is part of a well-established course with known credibility. The title accurately reflects the content, and the lecture successfully achieves its goal of bridging discrete and continuous dynamic programming. Overall, the lecture is of high quality, with strong scientific rigor and clear argumentation, though it could benefit from more concrete examples of solving HJB equations in practice.

The title accurately reflects the content, which focuses on continuous dynamic programming in the context of underactuated robotics.

8/10

Lecture from MIT OpenCourseWare, presented by an expert in the field, with rigorous mathematical derivations and references to course materials. The content is well-structured and aligns with established control theory.

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

• 00:00Introduction and recap of discrete dynamic programming
• 05:00Transition to continuous time and formulation of optimal control problem
• 10:00Derivation of the Hamilton-Jacobi-Bellman equation
• 15:00Discussion on the sufficiency and necessity of HJB
• 20:00Limitations of discrete methods and motivation for continuous approaches
• 25:00Challenges in representing value functions in continuous spaces
• 30:00Preview of algorithms for solving HJB equations
• 35:00Example of double integrator and non-differentiability of optimal cost
• 40:00Conclusion and outlook for future lectures

• Underactuated Robotics Course Website — Official course website with lecture notes, assignments, and additional resources.

• Underactuated Robotics Course Website — Course materials align with the lecture content.