inverted pendulum control with Q-learning (Reinforcement Learning)

I want to control an inverterd pendulum with Q-table. I simulated pendulum with simulink, but ''matlab function block'' doesn't work correctly. I want that (phi, phi dot, x, x dot) give to matlab fcn block,after it, action calculate and apply to pendulum . have any one this project or know about my problem? thanks a lot.

[Daniel Gomez]

I am having the same problem, I tried using a matlab function block but I get an algebraic loop error

[Canberk Suat Gurel]

Have you managed to solve this issue? If so can you explain how?

[Sara Hamdan]

I am having also the same problem. After a lot of trials, I found that we can't do RL in Matlab function block, so I did it in m-file and I called the Simulink file from the m-file. Did anyone try this? we can discuss the results.

[Emmanouil Tzorakoleftherakis]

You can now do RL both in MATLAB and in Simulink with Reinforcement Learning Toolbox. In Simulink you can use the RL agent block (no need to use a MATLAB Fcn block).

See an example here.