Hummingbird - Minigame of player competing against Reinforcement learning ML model
Gameplay: Player as Hummingbird competes against AI Hummingbird at absorbing the most nectar from multiple flowers
Vision: first-person view
Control:
- Mouse movement: PITCH and YAW
- WASD + Ctrl + Space: 3D translation
InstallationLog.txt is included for personal use to keep track of what was done. However it is set to public in case anyone encounters same issue to consult.
Note: The reason why the training environment is installed in WSL Ubuntu is that, there seems to be some dependency issues in Windows since ML-Agents Toolkit is not supported anymore. Training inside WSL would be limited since WSL2 would by default limit RAM usage to maximum only ~50% of the user's hardware.
(to check Memory and Swap space allocated to WSL, run free -h --giga within WSL distribution)
Some images:
CREDIT: ML-Agents course on Unity Learn platform, instructed by Adam Kelly