Hello,
I have been trying to use ROS publish and subscribe blocks for communicating between 2 different versions of MATLAB (R2018a and R2022a) on the same computer (same machine). The thing is, I am trying to run experiments on a hardware system that is compatible with R2018a or lower. However, I want to implement Reinforcement learning on the system. RL toolbox is available from R2019a onwards. Hence, I trained my RL agent using R2022a, and now want to deploy the trained agent on the hardware (which interfaces only through R2018a).
The system models, for the hardware interface and the RL agent are in Simulink. So the below description is for Simulink models in the respective MATLAB versions.
I was planning on using ROS publish blocks in R2018a (hardware compatible version) and corresponding subscribe blocks in R2022a (RL agent trained version), both versions on the same computer (my laptop). I thought this way, I will be able to communicate with the hardware using R2018a and recieve the states from the hardware system, which I can publish on a ROS topic. After this, I wanted the subscribe block in R2022a, to subscribe to that ROS topic and recieve the system states which can then be fed to the RL agent. This RL agent can then give out actions, which I planned to communicate the same way (in reverse manner) to the hardware. However, after a lot of tries and going through multiple solutions, I have not been able to figure this out. Any suggestions or ideas are welcome.
