how to create own environment in reinforcement learning

how to create own environment.

Answer 1

[Emmanouil Tzorakoleftherakis]

To create a MATLAB environment type

rlCreateEnvTemplate('myEnv')

This will create a template m file based on the pendulum system. You can use this as a starting point and make changes as needed for your application.

For Simulink environments, you can use blocks to represent the system dynamics and reward, and use the Agent block to train and simulate the agent.

[Kai Tybussek]

Is there some more documentation about custom environments exept this example? It´s not very detailed in terms of constraints for the different functions or properties. An Environment for a complete continuous Observation and Actionspace wld be very interesting to me, as well as an example on how to implement real hardware communication into an reinforcement learning algorithm. Like building a real cart-pole system and let the agent train in "real" scenarios.

best regards

Kai