MATLAB: How to change Agent policy option to continuing update the policy in Reinforcement Learning toolbox

Question

I using the toolbox example to generate Agent mex-code to deploy in Simulink,

and the show result only in static policy (turn learning off).

But, the reinforcement learning ebook denote user have the option of executing the static policy (turn learning off) or continuing to update the policy (turn learning on).

How can i change Agent policy option?

and, How should I do to deploy RL algorithm in target hardware.

the reinforcement learning example:

https://www.mathworks.com/help/reinforcement-learning/ug/water-tank-reinforcement-learning-environment-model.html

the reinforcement learning ebook:

https://www.mathworks.com/campaigns/offers/reinforcement-learning-with-matlab-ebook.confirmation.html

Answer 1

Hello,

Thank you for going over the RL ebooks! This point you mention is a general reference on what you can do after training your policy. Currently there is no out-of-the-box functionality that lets you do that in Reinforcement Learning Toolbox, so this is something you may have to implement yourself manually. This post may be helpful too.

Please note that training with/on hardware is a feature we are currently working on, so stay tuned!