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 for PITCH and YAW. WASD + Ctrl + Space for 3D translation
InstallationLog.txt is included for personal use to keep track of what was done. However it is set to public for anyone encountered 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 that well-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 this command "free -h --giga" within WSL distribution)
Some images:
CREDIT: ML-Agents course on Unity Learn platform, instructed by Adam Kelly