Reinforcement Learning training environments with ML-Agents Release 6.
- Replace following folders and files from the new released ML-Agents Project:
[offical ml-agents path]/Project/Assets→[UnityEnvs path]/Assets[offical ml-agents path]/Project/Packages→[UnityEnvs path]/Packages[offical ml-agents path]/Project/ProjectSettings→[UnityEnvs path]/ProjectSettings[offical ml-agents path]/Project/Project.sln.DotSettings→[UnityEnvs path]/Project.sln.DotSettings[offical ml-agents path]/com.unity.ml-agents→[UnityEnvs path]/Packages/com.unity.ml-agents[offical ml-agents path]/com.unity.ml-agents.extensions→[UnityEnvs path]/Packages/com.unity.ml-agents.extensions
- Modify
[UnityEnvs path]/Packages/manifest.json:- Remove the key
com.unity.ml-agentsor change the value of it from"file: ../../com.unity.ml-agents"to"file: ./com.unity.ml-agents" - Remove the key
com.unity.ml-agents.extensionsor change the value of it from"file: ../../com.unity.ml-agents.extensions"to"file: ./com.unity.ml-agents.extensions"
- Remove the key
Create a unique folder in [UnityEnvs path]/Assets/Environments and then implement your own training environment following the guide in ml-agents.