Back then, I had a setup of an agent consisting of 2 different brains, using a script to "toggle" between appropriate behaviours. Recently, I updated the ml-agents library to r3 and realised that Brain is no longer a scriptable objects. I am not sure I understand how to implement my model in the new setup. I have added 2 separate Behaviour Parameters components but that doesn't seem right. Should I "toggle" the Behaviour Parameters components as well, and how? Or should I implement the model differently? Any suggestion is welcome. Thanks!
There is an example with what you described. In: ML-Agents/Examples/WallJump/Scripts/WallJumpAgent.cs Look for the function Configure agent That example contains all the logic for that. SetModel("SmallWallJump", noWallBrain);