Hybrid reinforcement learning and traditional control environment

Hi,

I would like to setup a control system which uses the actuations dictated by an RL agent until some stopping condition is met and then for the remainder of the time use a traditional controller (i.e. PID etc).

Is this possible to setup using Simulink?

Answer 1

[Emmanouil Tzorakoleftherakis]

Hello,

You can put the RL Agent block in an enabled subsystem and use the desired condition to indicate when to use RL and when to switch to some other controller