From a technological point of view we are utilizing Dapr to create an easy to use Microservice through the gRPC transport layer. To put this into a diagram, we can think of this as the following:
The architecture exists out of a Server environment that is providing the Simulator integration while on the other side a Client is created through an SDK integration that allows us to write experiments easily. The entire gRPC layer is abstracted away from the end-user, such that the Subject Matter Expert (Data Scientist) is able to utilize this SDK in a way they are used to.
Find more information about the sdk here
To have pyglet correctly render as required in some environments (e.g. OpenAI), it is required to have a Display Server that can handle this. For Windows we utilize the XMing Server that can be downloaded from the following link: https://sourceforge.net/projects/xming/.
Once this is installed, we have to configure an environment variable that lets our WSL 2 environment know where this Server is running. Since WSL 2 is a full native linux system, we thus have to configure it to connect to our windows system.
For this fetch your local IP address and export it to the variable DISPLAY, e.g.: export DISPLAY="192.168.1.4:0".