Closed Gs1997XX closed 3 years ago
Hello! Appologies for the late response, I was having issues with my github account.
I haven't implemented this feature in the environment but I agree it would be useful! In theory you should be able to do so by following this notebook and replacing the hard coded moves in cell 18 with a keyboard input inside the stepping loop.
v3 (currently in beta here https://github.com/mdcrosby/animal-ai) outputs the agent position as an observation by default. You can use this to reconstruct the trajectory. Automatic output of the trajectory will be in one of the next releases.
@beyretb Hello, recently our team try to use your env with cite. We do not find the code to record the trajectory during the agent training?Is there a API to read the agent state and items state while playing the game?
Best wishes