Add AMDC tutorial article using Simulink Autogen

[noguchi-takahiro]

Context

Related Issues / PRs:

• New content for control using AMDC #85
• https://github.com/Severson-Group/AMDC-Examples/issues/6
• MathWorks documentation

We already have an article regarding Autogen (see this).

In this issue. there are several items that indicate that we want improvements:

Approach

1. Review this article
2. Address this comment
3. Create a new branch and create an article. Make it simple! Make sure to include a procedure how users can implement the Autogen
4. Create a PR by adding the article and to close the following issues:
   • Add AMDC tutorial article using Simulink Autogen #143
   • https://github.com/Severson-Group/AMDC-Examples/issues/6

[noguchi-takahiro]

@Daehoon-Sung I guess we can kick off this issue on Autogen article at this time? We can sync-up on it during lab hour on 11/14.

[mohamed-dek1]

Let use lab hours for reviewing all the material we already have, and come up with a plan for what we add to docs.amdc.dev.

[noguchi-takahiro]

Per 11/14, here is a plan of the article outline:

---

Simulink Automatic Code Generation for AMDC

• Explain an overview of the article
• The article should be simple. The idea of the article is that if AMDC users read the article, they can understand how to do Autogen from the beginning
• To do this, we need to document all required process

Example of Model Configuration

• Explain that we will use a simple integrator as an example
• Show block diagram with input, the integrator, and output

Procedure

Pre-Requisites

• User need to install dedicated MATLAB/Simulink toolbox/features (e.g., Embedded coder)

File Organization

• Provide a preferred file organization so that the AMDC can access the generated C-code

Model Setting

• Explain that we need to modify the Simulink setting to enable the Autogen (e.g., set Fixed-step)

Create a Reference Model

• Make a (discrete time) integrator
• Provide a process of how to create a reference model (see this)

Generate C-code

• Provide a process of how to generate C-code using Autogen feature, i.e., run slbuild(modelName.slx) command.

Integration with AMDC

• Provide an example C-code to call the Autogen files within SDK, i.e., we need a following code:

https://github.com/Severson-Group/ARL-eturbo/blob/1ae4479c934d486e06f233d98f9384fda36a545d/Embedded/My-C-Code/usr/bm_4dof/task_bm_4dof.c#L649

Results

• After running the AMDC, show the input and output value through logging feature

---

Next step

During next lab hour (on 11/21), we will demonstrate creating Simulink model to check if all process work. Meanwhile, we will take a note of the process, which will help us to create the article.

[noguchi-takahiro]

@elsevers do you have a chance to take a look at this comment, where a plan for the anti-windup article is available? I are planning to demonstrate the Autogen with Daehoon during lab hour on 11/21 to prepare for the article.

[noguchi-takahiro]

During lab hour on 11/21, we updated MATLAB to be R2025b in the ME-ELEV-08, including the following required tool boxes:

• Simulink
• Embedded coder
• Simulink coder

Let us continue this in the next lab hour (12/5).

[noguchi-takahiro]

Per 12/5 lab hour, we are not able to do this due to this issue. We are aiming to go through this issue on 12/7.

[mhmokhtarabadi]

I followed this article to create c-code from my simulink model. But there were some settings I needed to change. See this comment.

[noguchi-takahiro]

1. Open MATLAB
2. Save a new .m file as setup.m
3. Open Simulink
4. Click Blank model
5. Double click an area in Simulink to add Discrete-time integrator.
6. Let us make a continuous time transfer function as a Plant (= 1)
7. Add rate transition block before the integrator.
8. In the rate transition, put Ts as a sampling time.
9. Add the rate transition, put -1 as a sampling time.
10. In setting, Solver-> Solver Selection -> Type: Fixed-step
11. Code Generation-> Browse -> select ert.tlc Embedded coder
12. Model Settings / Code Generation / Build process / Generate code only ---> check
13. Model Settings / Code Generation / Optimization / Target specific optimizations / Leverage target hardware instruction set extensions ---> None
14. Model Settings / Code Generation / Templates / Custom templates / Generate an example main program ---> Uncheck
15. Click OK
16. In m file, define Ts = 1/(10e3), Tsim = Ts/10.
17. Select the integrator, and right click -> select Create Subsystem from Selection.
18. Right click on the subsystem - > Block parameters (Susbsystem) and check Treat as atomic unit. Clock OK.
19. Right click on the subsystem -> Subsystem & Model Reference -> Convert -> Reference Model
20. In New model name: integrator
21. Click Apply and Convert
22. Rename the reference model to be integrator
23. Double Click and Model Setting -> Reference Model Setting
24. In Solver -> Solver details, Put Ts.
25. Copy the following code in .m file:

%% Autogen code for the controller
model='integrator'; % Name of the controller to be built
slbuild(model);     % Generates the autogen code
oldFolder = cd('C:integrator_ert_rtw\');
command = 'for /r %i in (*.c, *.h) do copy /y %i ..\autogen';
[status, cmdout] = system(command);
cd(oldFolder);

@Daehoon-Sung will work on making the article within the next week and aim to merge it in.

[Daehoon-Sung]

I need to add screenshots to make the article, and I am planning to finish the article today.