Closed kwea123 closed 5 years ago
Also, for the actions, can you provide information of what values correspond to what actions? For example the first dimension, does 1 = move forward or 2 = move forward? Same for other actions.
If the above information can be summarized into a table, that would be a huge help. Thanks!
I guess all the information you mentioned above is a part of the challenge itself.
Hi @kwea123,
What each of the actions do is described in cell 3 of the jupyter notebook: https://github.com/Unity-Technologies/obstacle-tower-env/blob/master/examples/basic_usage.ipynb.
The specifics of how those actions impact the environment are something we'd like people to play with and discover themselves. They also vary depending on the presence of obstacles or things like whether the agent is already in motion or not, and are therefore not simple to enumerate.
Thanks all. Closing the issue.
For example, moving one step forward is 0.1m; rotating the camera one step is 18 degrees; and the size of the maze (for one block) is 3mx3m... Is there such a table that we can refer to?