Advanced methods in Control and Optimization, SoSe-2022, RWTH Aachen

This repository is is only for referring to my work for the Seminar Course named "Advanced methods in Control and Optimization" held at RWTH Aachen, offered by "Chair of Intelligent Control System" organised by "Mr. Dennis Gramlich".

In this seminar, advanced methods in the area of control and optimization are studied. In particular, topics in optimal control, nonlinear control, robust control, and stochastic control.

Goals:

1. To be able to search, read and understand international, English language research literature.
2. Understand and explain research problems and their solution.
3. Present the state of the art of research in written and oral form in English.
4. Relate the big picture to the details of a technical topic.

Deliverables (all in English):

1. Paper on the chosen topic (6-8 pages excl. references).
2. Presentation of 45 minutes plus 15 minutes Q&A.

You have to hand in a six-page review paper that contains:

1. Some background on the broader topic of your paper, incl. helpful references.
2. The state of the art on the more specific problem, incl. helpful references.
3. The main contributions and methods used to solve the problem.
4. A discussion of the presented results.
5. The paper should be written using LATEX.

Assigned Paper:

1. Title: Synthesis and stabilization of complex behaviors through online trajectory optimization.
2. Authors: Tassa, Yuval and Erez, Tom and Todorov, Emanuel
3. Link: https://homes.cs.washington.edu/~todorov/papers/TassaIROS12.pdf
4. Supervisor: Dennis Gramlich
5. Synonpsis: The paper is about an online trajectory optimization method and software platform applicable to complex humanoid robots performing challenging tasks such as getting up from an arbitrary pose on the ground and recovering from large disturbances using dexterous acrobatic maneuvers.

Brief Tasks:

1. Derive Differential Dynamic Programming (DDP) from the Bellman principle.
2. Present DDP in a concise and easily understandable form.
3. Proof the (local) quadratic convergence of DDP.
4. Introduce line search and regularization techniques to improve the convergence properties of DDP.
5. Implement Differential Dynamic Programming for an example.

Contents of the Repo:

1. Latex project folder for the presentation and the report aling with separate pdfs.
2. Implementation of DDP using JAX in Python.