Issue with getting started with bluesky

[r-ejilemele]

I'm working on using bluesky to visualize a multiagent reinforcement learning scenario along with @Jaroan. I was having a little trouble getting the bluesky python library started. To explain what I wanted to do, I wanted to spawn agents in an area of set width and height and then move them by by certain amounts and in certain headings every time step. And then these movements would be displayed through bluesky's gui after the code has finished running. I was having a bit of trouble looking through the documentation to get started.

[jooste]

Hi @r-ejilemele,

Welcome to the BlueSky community. Moving aircraft by a certain amount each time step is indeed in a nutshell what BlueSky does 😉. Combining BlueSky with an RL library can be done in two ways:

1. you use the RL library (or your own code) to maintain an event loop. In your loop you call bluesky.sim.step() to progress the simulation state. Interaction with the simulation can be done via stack commands (bluesky.stack.stack()) or by directly interacting with the BlueSky objects. If you take this approach you'll need to do your own visualisation, or you record all commands given to the simulation into a scenario file that you can then afterwards play back with a regular instance of BlueSky.
2. You run BlueSky normally, and implement your RL agent in a BlueSky plugin. In this case, BlueSky controls the event loop. Via your plugin you interface with your RL model to retrieve new commands per time step.