[Enhancement] Add Doom-Env - Doom Slayer goes to OpenEnv School to learn RL #228

AlirezaShamsoshoara · 2025-11-26T17:50:15Z

Add Doom Environment with ViZDoom Integration

⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣠⢾⠍⡉⠉⠙⣿⣆⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣠⣴⠾⠿⠽⢷⣶⣤⡀⠀⠀⠀⠀⠀⠀⠀⢀⣟⡟⣠⣿⣶⡀⣷⡻⡄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢠⠴⡟⡋⡀⠀⣀⣀⠀⠀⠉⠛⣦⡀⠀⠀⠀⠀⠀⠀⢿⣅⣽⣿⣿⣷⣿⣿⠃⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⡴⠃⢀⢾⣿⣿⣿⣯⣬⣽⣿⣀⡀⠈⠙⣆⠀⠀⠀⠀⢀⣸⣯⣿⣾⡷⢻⣿⠋⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣜⢁⠁⣾⡿⣙⠿⣯⣭⣍⣹⠼⠋⠁⣴⠀⢘⣧⠀⠀⡴⢛⣭⢟⠽⠋⢠⣼⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⢻⠘⢸⡿⢷⣬⣧⡀⠀⠀⠀⢀⣤⠾⢿⡇⠘⣿⡆⣸⠛⣿⡿⣟⡀⠀⡾⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣾⣦⣿⣿⡄⠈⢿⢿⣷⣶⡾⠋⠁⠀⣸⠇⡰⠛⢷⣷⣻⡿⠺⣿⣿⠽⠋⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣀⣀⣀⣀⣀⠀⣴⠏⠀⣿⠙⢻⣿⣄⠈⠀⠸⠀⠉⠀⣠⣾⠟⢀⣧⡇⠀⢽⣿⣿⣬⣼⣿⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣰⠚⣿⣿⣿⣿⣿⡿⠟⢛⣰⣿⣧⣷⣝⡿⣷⣞⢷⣄⣲⣾⣿⡃⢰⡿⡟⢀⣴⣿⣿⣿⣯⡿⠿⣿⣶⣤⣀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣠⠞⢁⣼⣿⣿⣿⡟⠋⠁⣉⣽⣿⣿⣿⣿⣿⣽⣯⣿⡄⠉⠁⢷⣬⣹⣿⣿⣤⡾⠁⣸⣿⣿⡟⠁⠀⠀⢹⣿⣿⣷⡆⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣾⡷⠞⣫⣾⣿⣿⣿⣧⡀⣤⠀⠈⣻⣿⣿⣿⣿⣿⣿⣿⣷⣖⠀⠘⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⠃⠋⠻⢤⣅⡺⢦⡀⠳⡄⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⣰⣿⣯⣴⠞⠁⣀⣿⣿⣿⣿⣷⣄⣤⠤⢊⣿⣿⣿⣿⣿⣿⣿⣿⣯⣴⣴⣶⣿⣿⠟⣸⣿⣿⣿⣿⡏⡆⠀⢠⣤⣠⣥⠀⡟⣶⣿⡄⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⣽⡿⣿⣏⡀⠀⠹⣟⣿⡿⣿⣿⣋⣶⣺⡽⣿⣏⣅⠛⠂⠴⠶⠿⠿⠃⠈⠉⠻⣷⣶⣿⣿⣿⣿⡿⠀⣿⡄⠈⣷⣮⠙⢀⡿⠘⢻⣇⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⢸⣧⢻⡶⠀⠀⠘⢿⣿⣿⣿⣿⣿⠋⠉⠀⠀⠉⠻⠿⠶⠶⠶⠦⠴⠞⠛⠷⠗⠈⠛⢿⣿⣿⡿⢁⣼⠯⠄⠀⠀⠀⣠⡞⠁⣠⣾⣿⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⢸⣻⣾⣷⡀⢐⠀⣿⣿⣿⣿⣿⠁⠀⠠⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⡁⠰⣾⣿⠀⠀⠈⢻⣿⣅⢿⣇⠀⠀⠀⠀⢀⣿⡟⠀⡷⢿⢿⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠙⣿⡎⣤⣌⡰⣿⣿⣿⣿⣟⠀⠠⠀⢀⡀⠀⠂⠀⠀⠉⠉⠉⠈⠉⠙⢾⣭⡤⠂⠀⠀⠹⣿⣎⣿⣶⣒⣿⣷⣿⣯⣮⡵⣿⣾⣿⡀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⢻⢿⠛⣿⠛⠛⢿⣿⣿⣃⢀⣀⣀⠀⣀⣤⣾⠓⠶⠖⠷⣤⣄⡀⠀⠀⠀⠀⠀⠀⢠⣿⣿⣾⣿⣿⣿⠍⣩⣉⣿⡆⠰⣿⣭⡇⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⡸⠾⢴⡇⠀⠀⢸⣿⣿⣯⣭⣿⣿⣿⡿⠛⠛⠛⠛⠛⠛⠛⠟⠻⣷⣶⣴⣶⣮⡴⠫⢾⣿⣿⣟⠉⣹⣿⣿⣿⣿⣷⣄⠸⢿⡇⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⣴⠋⡽⠁⡾⠀⠀⠀⣼⣿⣧⣁⣴⣶⠾⢿⣿⡶⠀⠒⠒⠂⠀⠀⠀⣰⣾⣧⣌⣉⠙⠂⢠⢿⣿⣿⣫⡿⠿⠋⠉⠈⠙⢻⣽⢧⠀⣽⣄⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⢀⣴⡻⣻⣼⣿⣰⠇⠀⠀⠀⠉⣁⣿⣟⢉⣼⣶⣶⡿⠿⣿⡟⠛⠛⠛⣷⣾⢿⣯⣤⣤⡉⠳⡶⢋⡞⣿⣿⣇⠀⠀⠙⠀⠀⠀⢀⣿⣫⠇⣈⣁⣣⡀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⣠⠎⠉⣰⣿⣿⣿⠉⢲⣤⠀⠀⠾⣿⣿⣳⣜⢿⡟⠫⢠⣶⣾⣷⣤⣤⣼⣯⡤⣤⣀⣻⠻⣿⣦⣠⠞⣼⣿⡿⢿⣤⡸⣷⣦⣤⣴⣿⣿⣯⠼⢥⣈⣿⡗⠶⢤⡀⠀⠀
⠀⠀⠀⠀⢰⡃⠀⣼⣿⣿⣿⣿⣷⣤⣁⣀⣤⣾⣿⣿⣿⣿⣿⣿⠷⣾⣟⣀⣫⣄⣀⣀⣠⣄⠘⢿⡤⠴⣷⡿⠃⠘⡽⣿⣃⠘⣿⣿⣿⣿⣿⣿⣿⠿⡿⠟⠀⠘⣝⢿⡆⠀⠻⣦⡀
⠀⠀⠀⢀⡏⢀⣾⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⡿⢳⣿⣯⣿⣿⣿⡿⣾⣿⡏⠙⣿⡉⠙⡍⠉⢿⣟⣴⠶⠾⢿⣟⠷⣿⡿⠿⣷⣾⡇⢻⣿⣿⡏⡴⠞⠻⣞⡍⠙⢫⣿⣧⠀⠀⠘⠃
⠀⠀⢀⣾⢿⡾⢷⣿⣿⠋⣿⣿⣿⣿⣿⣿⣿⡗⣼⢿⣿⣿⠘⣿⣿⣷⣼⣣⣶⠾⠿⠛⠶⣦⠚⣠⣴⣿⣿⠋⢰⣼⣯⠁⠐⢺⣿⣿⣮⣿⣿⣿⡟⠂⢀⣽⡓⡀⠒⢹⣿⠇⠀⣤⡀
⠀⠀⣿⠿⣾⣳⣼⣏⠛⠛⢿⣯⣶⣿⢋⣼⣿⢱⣟⣷⣮⠻⣷⠘⠿⣿⣭⣉⡉⣠⣤⣤⣄⣉⣉⣁⣾⡿⠟⣠⣾⡏⣡⠎⠀⢸⣿⡌⣿⣿⣿⣿⣟⡂⠠⢿⡅⢨⡏⣾⣟⠀⠀⠈⠁
⠀⢸⡿⠓⢀⣿⣿⣿⡷⣦⣼⠟⣹⡵⠛⢳⢟⣾⣿⣿⡿⠀⣿⠄⡀⣿⣯⠙⣿⡟⠛⢛⠛⣿⣿⡏⢉⡇⠀⢯⣿⡇⡅⢴⠀⢸⣾⡇⠸⣟⠹⣿⢿⡏⢰⣿⣆⣈⠁⣽⣿⠀⠀⠀⠀
⢀⡖⠘⠃⢠⣿⣯⡟⠻⣿⣻⡟⠃⠀⠀⠸⣿⢿⣿⣿⣿⣾⣿⣿⣿⣿⣿⣧⢸⣧⣤⣭⣤⣿⣿⡔⢿⣿⡿⣿⣿⣿⣷⣤⣠⣿⣿⠃⠀⣿⣇⣿⣿⣷⣿⣿⠿⢽⣷⣩⣿⠀⠀⠀⠀
⣾⠁⣠⠹⣿⣿⡟⠻⣶⣿⢻⡇⠀⠀⠀⠀⠈⢹⡿⣿⣿⣿⣿⢟⣟⢿⢿⣿⣿⡷⠶⠶⠶⠈⢯⡻⡄⢻⣿⢀⠙⢿⣿⣿⣷⡟⠁⢀⣴⢟⣺⣿⣿⣿⣥⣽⣶⣄⣈⣿⣿⠀⠀⠀⠀
⣭⠎⠿⢠⡟⢿⣿⣷⣽⣿⣼⡇⠀⠀⠀⠀⢠⣿⢿⡛⢿⡿⣿⡾⣿⡇⢠⣿⣿⡇⠀⠀⠀⠀⣈⢻⡖⢸⣿⢿⣾⢏⠟⠛⢿⣧⣀⣸⣴⡿⢻⣿⣻⣍⠉⣉⠛⣛⠛⠛⢿⡷⠀⠀⢀
⢳⣶⠖⠈⢿⣿⣛⠹⣿⣿⢸⡃⠀⠀⠀⣠⠟⣩⠞⠀⠈⣿⡟⣵⡿⠃⣼⣿⣿⠁⠐⠀⠘⠃⠉⣸⣇⠀⠹⣦⢻⣟⠀⠀⠀⠹⣿⣴⣯⣼⣿⣿⣿⣿⡄⣿⡀⢿⣰⡇⢸⡇⠀⠠⠋
⠸⣹⡶⠀⢸⣿⣿⣿⣷⣛⢻⡇⠀⠀⢠⡷⠃⠁⠀⠀⠀⣿⠸⣿⠀⠠⣿⣿⣧⡀⠀⠀⠀⠀⢰⣿⣄⠁⠀⣹⢦⣿⣦⠀⠀⠀⣿⣿⣿⡏⣿⡏⡛⠟⢲⣶⢶⣾⣷⡭⣸⡴⠊⠀⠀
⠀⢹⡄⣄⡘⣿⣿⣿⣿⠹⡿⠁⠀⠀⣿⠇⠀⠀⠀⠀⡶⠘⡇⣿⡃⠂⣻⣿⣿⣷⡄⠀⠀⠀⢸⣿⣝⡓⢰⣿⣾⡏⣿⣦⠀⠀⢹⣾⣿⡎⢰⣷⣓⠀⣼⣿⢸⣿⢹⡆⢿⠇⠀⠀⠀
⠀⠀⠙⠻⣿⣿⠧⠭⠭⠟⠁⠀⠀⣸⡽⢐⠀⠀⠀⢸⣇⣸⡷⣿⠃⠀⢿⣿⣿⣿⣿⣿⣷⣦⣿⣿⣯⡟⢺⣿⣿⣇⣸⡿⡇⠀⢀⡟⣿⡧⢸⣷⡌⢀⣿⣿⣼⣿⠮⣿⠋⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠈⠁⠀⠀⠀⠀⠀⢀⣟⣷⡿⠀⠀⠀⠀⠉⢸⡇⡷⠀⢀⠈⠻⣿⣿⢿⣿⢿⣿⣿⣿⣿⣷⣾⣿⣿⡇⠉⠀⠀⠀⢸⡇⢸⣿⡾⡿⣧⣼⣿⠵⣿⣇⡾⠁⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣾⣼⡛⢸⡄⠀⠀⠀⢸⣧⢳⣀⠀⠀⠀⣿⢋⡟⠈⢧⢻⣿⣿⣿⣿⣿⣿⣿⣷⡀⠀⠀⠀⠈⡇⠀⢯⡇⠀⠉⠙⠙⠉⠉⠋⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣿⣿⢹⡸⣷⠀⠀⠀⠄⠻⢷⣄⣀⢀⣼⣣⠟⠀⠀⠈⢣⠹⣿⣿⣿⣿⣿⣿⠿⢷⡄⠀⠀⠀⠀⠀⢸⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣿⣿⡈⢁⣽⣷⡆⠀⠀⠀⠀⢈⣽⣿⡿⠃⠀⠀⠀⠀⠀⠙⣌⢻⣿⣿⣿⣿⠀⠈⢿⣦⠓⠀⠀⠀⣸⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣸⠚⢿⣷⣿⣿⣯⣻⡄⠀⠀⢀⣾⠟⡿⠁⠀⠀⠀⠀⠀⠀⠀⠈⢦⡻⣿⣿⣷⡀⢠⣾⣫⡿⣬⡃⠆⠛⣧⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢠⣻⣷⢾⣟⠛⠁⠉⢻⣿⣆⣠⡾⢿⣿⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠘⣇⣸⣿⣿⣾⣾⣿⠇⠀⠈⢙⡟⠿⢻⣆⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣿⣫⡾⠿⣦⣀⣀⣠⡿⢿⣏⡴⣿⡏⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣿⣿⣿⣿⣿⣿⣤⣤⣤⡞⠁⢂⣹⣿⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣸⣉⣿⡳⠀⠀⠈⠁⠀⠀⠈⢿⡄⣿⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠘⣿⢿⣿⣿⣿⣷⣄⠀⠀⠀⢀⡀⠘⣿⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣰⣷⣻⣿⡟⠶⠶⠤⠤⠀⠀⠀⣸⣿⡿⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠙⣿⣿⣿⣿⣿⣿⣶⣖⣾⠭⡁⠈⢿⣳⣦⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⢰⣟⣿⣽⠋⣿⠃⢤⣭⣭⠀⠀⠀⣠⣟⣿⣆⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣾⡿⣿⣿⣿⣿⣯⡥⠶⠀⣛⠀⢶⡿⡬⣷⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⣶⡿⢱⣿⣰⡿⠿⠶⢭⣦⠀⠀⣰⡿⢁⢿⣾⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢰⣿⣢⣿⣿⣿⣿⣷⣶⠖⠛⠙⢷⣌⡉⠹⣷⡇⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⢰⣿⣗⣿⣿⣿⡀⣶⣶⠀⣹⣷⣾⣿⣷⡼⣯⣿⣆⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣾⢿⣿⣿⣿⣿⣿⣿⣿⣗⣼⠄⠀⣿⣷⣀⣿⢷⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠈⢹⣿⡿⢻⣿⣷⣽⣏⣰⣿⣿⣿⣷⣶⣧⢹⣷⡟⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠸⣭⠿⣿⣿⣿⣿⣿⣿⣿⣿⣥⣤⣾⠏⠻⡇⣿⡏⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⢸⣿⢿⣿⣿⣟⠻⠿⠿⠛⠹⣿⣿⣿⣿⣾⣿⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⡇⣻⣿⣿⣿⣿⣿⡍⠛⠛⠋⠁⠀⣀⢿⣿⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⢸⡿⣿⣿⣿⣿⡗⠓⣤⠀⢀⣤⣿⢃⣸⠟⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠘⢿⠻⣿⣿⣿⣿⣿⣏⣹⣧⢀⣾⡉⢸⣿⡆⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⣼⡿⣿⣿⣿⣛⠁⠘⡋⠙⣋⣥⣿⢾⡏⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢣⡜⣿⣿⣿⣿⡏⠛⢩⡉⠀⡛⢸⣿⢷⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⢀⣿⣷⣿⣿⣿⣯⣤⡤⡒⣛⣭⣭⢾⡿⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⣿⢹⣿⣿⣿⣿⡦⠼⣷⠚⣩⠏⠹⣯⡀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⣿⡅⢠⣿⣿⣿⣧⣤⠾⠟⢛⣫⡵⣿⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣳⢿⣿⣿⣿⣿⡶⣿⡞⠋⠀⠀⢻⣧⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⢰⣿⢡⢿⣿⣿⣿⣀⡀⠚⣠⣼⠁⢀⣿⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣇⣼⣿⣿⣿⣿⣇⣼⣷⣶⣿⠟⠀⣿⣇⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⣞⠇⣶⣷⣬⣭⣉⣛⢛⣛⠉⣩⡷⢾⣿⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⡯⢽⣿⣿⣿⣿⣟⣉⣩⣤⡤⠶⠂⠸⣾⡄⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⣾⣿⡷⣿⣽⣾⣟⣿⣭⠈⠁⠀⣿⣠⣼⣿⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣷⢦⣿⣿⣿⣿⣿⣯⣁⣾⣷⣶⣿⠣⣷⣵⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⢀⣾⢻⠤⠟⠓⠚⠻⢧⣀⠀⠀⠀⠙⣿⣿⣯⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⢻⣿⣿⢡⠿⠛⠋⠉⠩⠀⠀⠀⠒⠄⠞⣦⠀⠀⠀⠀⠀
⠀⠀⠀⠀⢸⣷⣴⢞⣏⣀⠀⡀⠀⣹⣦⠾⠟⢂⡍⠻⣷⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢨⠿⢛⣿⡶⢷⣤⣄⣀⣀⡀⢠⣴⣀⠠⡼⣿⡁⠀⠀⠀⠀
⠀⠀⠀⠀⢘⡃⢰⠀⡀⠀⠀⢀⡀⠀⠀⠈⠀⠀⢩⠈⣽⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢸⣼⣿⣿⡀⣀⠒⣿⣿⠇⠀⠀⠀⠀⡀⡇⠘⣿⡀⠀⠀⠀
⠀⠀⠀⠀⠀⡇⢸⠈⠁⠀⠀⢸⡇⠀⡇⠀⢖⣔⣾⣾⡋⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠑⣟⢿⣀⣿⣾⢿⣿⡇⠈⠁⠀⠀⠀⣿⣤⣧⠇⠀⠀⠀

Summary

This PR introduces a new Doom environment for OpenEnv, wrapping the ViZDoom platform to provide visual reinforcement learning capabilities for Doom-based scenarios. The environment supports multiple scenarios, configurable resolutions, discrete/continuous action spaces, and includes comprehensive documentation with visual examples.

Overview

The Doom environment (doom_env) integrates ViZDoom - a Doom-based AI research platform - into the OpenEnv framework, enabling agents to:

Learn from visual observations (RGB or grayscale screen buffers)
Execute actions in 3D game environments
Train on multiple built-in scenarios (basic, deadly_corridor, defend_the_center, etc.)
Track game variables (health, ammo, kills)
Deploy via Docker or run locally

What's Included

Core Implementation

models.py - Data models for actions and observations
- DoomAction: Action dataclass with support for discrete actions or button combinations
- DoomObservation: Observation dataclass with screen buffer, game variables, rewards, metadata
client.py - HTTP client for connecting to Doom servers
- DoomEnv: Full-featured client with rendering support (OpenCV/matplotlib)
- Automatic numpy type conversion for JSON serialization
- Client-side rendering with render() method
server/doom_env_environment.py - Core ViZDoom wrapper
- Configurable scenarios, resolutions, screen formats
- Discrete actions and button combinations support
- Episode management and state tracking
server/app.py - FastAPI server application
- Environment variable configuration (scenario, resolution, format)
- Web interface integration
- Health check endpoint

Docker Support

server/Dockerfile - Standalone Docker image
- Based on python:3.11-slim
- Includes all ViZDoom system dependencies
- Configurable via environment variables
- Works for both local builds and HuggingFace deployment
- Build command: docker build -t doom-env:latest -f src/envs/doom_env/server/Dockerfile src/envs/doom_env

Documentation

README.md - Comprehensive environment documentation
- Quick start guides
- Scenario gallery with descriptions
- API reference
- Configuration options
- Deployment instructions
- Visual examples with ASCII art Doom Slayer
GIF_GENERATION.md - Guide for generating scenario GIFs
- Step-by-step instructions
- Example commands
- Troubleshooting tips
TEST_PLAN.md - Comprehensive test strategy (future implementation)
- 67 planned test cases
- Test categories and fixtures
- Success criteria

Utilities

generate_gifs.py - Script to generate scenario visualization GIFs
- Supports all ViZDoom scenarios
- Configurable steps, resolution, FPS
- Automatic output to assets/ directory
example.py - Example usage script (in examples directory)
- Demonstrates basic usage
- Docker and local modes
- Rendering examples
doom_visualizer.py - Real-time game visualizer (in examples directory)
- OpenCV-based visualization with keyboard controls
- Matplotlib fallback
- Auto-scaling for different resolutions
- Interactive controls (arrows, space to shoot)

Assets

assets/doom_slayer_at_openEnv_school.png - Custom Doom Slayer artwork
assets/README.md - Assets directory documentation

Key Features

1. Multiple Scenarios

Basic - Simple target shooting (beginner-friendly)
Deadly Corridor - Navigate corridor while avoiding fireballs
Defend the Center - Survival mode defending against monsters
Defend the Line - Protect a line from advancing enemies
Health Gathering - Navigate maze collecting health packs
My Way Home - Navigation to goal location
Predict Position - Predict enemy positions
Take Cover - Strategic cover-based combat

2. Flexible Configuration

Environment Variables:

DOOM_SCENARIO=basic                 # Scenario selection
DOOM_SCREEN_RESOLUTION=RES_640X480  # Resolution (160x120 to 1024x768)
DOOM_SCREEN_FORMAT=RGB24            # RGB24 or GRAY8
DOOM_WINDOW_VISIBLE=false           # Show game window
ENABLE_WEB_INTERFACE=true           # Enable /web UI

3. Action Spaces

Discrete Actions: Simple integer action IDs (0-3 for basic scenario)
Button Combinations: Full control with custom button lists

4. Rendering Options

Web Interface - Browser-based UI at /web endpoint
Client-side Rendering - OpenCV or matplotlib visualization
RGB Array Mode - Return numpy arrays for custom processing

5. Docker Deployment

Local Build:

docker build -t doom-env:latest -f src/envs/doom_env/server/Dockerfile src/envs/doom_env
docker run -p 8000:8000 -e DOOM_SCREEN_RESOLUTION=RES_640X480 doom-env:latest

HuggingFace Deployment:

cd src/envs/doom_env
openenv push

Technical Details

Architecture

Client-Server Model: HTTP-based communication via OpenEnv framework
ViZDoom Integration: Native Python bindings to Doom engine
Screen Buffer Format: Flattened RGB/grayscale arrays for efficient transmission
State Management: Episode tracking, step counting, game variables

Fixed Issues

Docker Build for HuggingFace
- Fixed build context to work with both local and HF deployments
- Changed from copying src/core/ to installing openenv-core via pip
- Updated paths: WORKDIR /app/env, COPY . ., pip install -e .
Environment Variable Configuration
- server/app.py now reads DOOM_* environment variables
- Docker resolution changes take effect at runtime
- Proper defaults for all configuration options
JSON Serialization
- Added numpy type conversion in client.py::_step_payload()
- Handles np.int64, np.float32, numpy arrays
- Filters out None values from payload
Rendering Window Size
- doom_visualizer.py auto-scales windows based on resolution
- Uses cv2.resize() with INTER_NEAREST for pixel art preservation
- Target window size: 1024px width with appropriate scaling
OpenEnv Validation
- Restructured server/app.py to follow snake_env pattern
- Added main() function with proper if __name__ == "__main__" block
- Passes openenv validate for multi-mode deployment

Dependencies

Python Packages (from pyproject.toml):

dependencies = [
    "openenv-core>=0.1.0",
    "fastapi>=0.115.0",
    "pydantic>=2.0.0",
    "uvicorn[standard]>=0.24.0",
    "requests>=2.31.0",
    "vizdoom>=1.2.0",
    "numpy>=1.19.0",
]

Optional:

opencv-python>=4.5.0 - For client-side rendering
matplotlib>=3.3.0 - Rendering fallback
imageio>=2.9.0 - For GIF generation

System Dependencies (for ViZDoom):

build-essential, cmake
libboost-all-dev
libsdl2-dev, libfreetype6-dev
OpenGL libraries (libgl1-mesa-dev, libglu1-mesa-dev)

File Structure

doom_env/
├── __init__.py                           # Module exports
├── models.py                             # DoomAction, DoomObservation
├── client.py                             # DoomEnv HTTP client
├── README.md                             # Main documentation
├── GIF_GENERATION.md                     # GIF generation guide
├── TEST_PLAN.md                          # Test strategy
├── openenv.yaml                          # OpenEnv manifest
├── pyproject.toml                        # Dependencies
├── uv.lock                               # Locked dependencies
├── generate_gifs.py                      # GIF generation script
├── assets/                               # Generated GIFs and images
│   ├── doom_slayer_at_openEnv_school.png
│   └── README.md
└── server/
    ├── __init__.py
    ├── doom_env_environment.py           # ViZDoom wrapper
    ├── app.py                            # FastAPI server
    └── Dockerfile                        # Docker image

examples/
├── doom_example.py                       # Basic usage example
└── doom_visualizer.py                    # Interactive visualizer

Usage Examples

Basic Usage

from doom_env import DoomEnv, DoomAction

# Connect to server
client = DoomEnv(base_url="http://localhost:8000")

# Reset environment
result = client.reset()
print(f"Initial health: {result.observation.game_variables[0]}")

# Take actions
for _ in range(100):
    action = DoomAction(action_id=1)  # Move left
    result = client.step(action)

    if result.observation.done:
        print(f"Episode finished! Total reward: {result.reward}")
        break

client.close()

Docker Mode

from doom_env import DoomEnv, DoomAction

# Start container automatically
client = DoomEnv.from_docker_image("doom-env:latest")

result = client.reset()
result = client.step(DoomAction(action_id=0))

client.close()

With Rendering

client = DoomEnv.from_docker_image("doom-env:latest")
result = client.reset()

for _ in range(100):
    result = client.step(DoomAction(action_id=1))
    client.render()  # Display the game

client.close()

Visual Examples

The environment includes a custom Doom Slayer ASCII art and supports generating GIFs of all scenarios:

python generate_gifs.py basic --steps 500 --resolution RES_640X480

Testing

Comprehensive test plan covering:

15 Model Tests - Data validation, serialization
18 Environment Tests - ViZDoom wrapper functionality
20 Client Tests - HTTP client, rendering, serialization
14 Integration Tests - End-to-end, Docker, performance

Note: Test implementation deferred to future PR

Documentation Updates

Added Doom environment card to docs/environments.md
HuggingFace Space: https://huggingface.co/spaces/Crashbandicoote2/doom_env
Comprehensive README with examples and API reference
GIF generation guide for creating scenario visualizations

Validation

Passes openenv validate for multi-mode deployment
Docker builds successfully: docker build -t doom-env:latest -f src/envs/doom_env/server/Dockerfile src/envs/doom_env
Runs locally: python -m doom_env.server.app
Deploys to HuggingFace: openenv push works correctly
Al environment variables respected at runtime
Cient-server communication verified with example scripts

Related Issues

Implements visual RL environment support for OpenEnv
Adds multi-scenario support with 8 built-in Doom scenarios
Add examples for RL training in the example folder

Future Work

Implement comprehensive test suite
Add support for custom WAD files
Implement multi-agent scenarios
Add more advanced scenarios (deathmatch, CTF)

Links

HuggingFace Space: https://huggingface.co/spaces/Crashbandicoote2/doom_env
ViZDoom Documentation: http://vizdoom.cs.put.edu.pl/
ViZDoom GitHub: https://github.com/mwydmuch/ViZDoom

Breaking Changes

None - This is a new environment addition.

Checklist

Code follows project style guidelines
Documentation is comprehensive and clear
Docker builds successfully (local and HuggingFace)
Environment validates with openenv validate
Example scripts provided and tested
README includes usage examples
Added to environments documentation
HuggingFace Space deployed and working

…roperly

… verifcation

AlirezaShamsoshoara · 2025-11-26T17:51:49Z

The HF Space for the pushed doom-env is available here:
https://huggingface.co/spaces/Crashbandicoote2/doom_env

AlirezaShamsoshoara · 2025-11-26T17:53:43Z

Adding reviewers here @init27 @Darktex @HamidShojanazeri @jspisak

.github/workflows/docker-build.yml

Copilot

Pull request overview

This PR adds a comprehensive Doom environment to OpenEnv by integrating ViZDoom, a Doom-based AI research platform for visual reinforcement learning. The implementation provides visual observations, multiple built-in scenarios, flexible action spaces (discrete/continuous), and comprehensive tooling for development and deployment.

Key Changes:

Complete ViZDoom wrapper with 8 built-in scenarios (basic, deadly_corridor, defend_the_center, etc.)
HTTP client-server architecture with FastAPI backend and Python client
Docker deployment support with configurable environment variables
Comprehensive test suite (65 tests across models, environment, client, and integration)
Visualization tools (web interface, OpenCV/matplotlib rendering)
Documentation with scenario gallery, deployment guides, and troubleshooting

Reviewed changes

Copilot reviewed 28 out of 39 changed files in this pull request and generated 18 comments.

Show a summary per file

File	Description
`src/envs/doom_env/models.py`	Data models for actions and observations with screen buffer support
`src/envs/doom_env/client.py`	HTTP client with rendering capabilities and numpy type conversion
`src/envs/doom_env/server/doom_env_environment.py`	Core ViZDoom wrapper implementing OpenEnv Environment interface
`src/envs/doom_env/server/app.py`	FastAPI application with environment variable configuration
`src/envs/doom_env/server/Dockerfile`	Standalone Docker image with ViZDoom system dependencies
`src/envs/doom_env/pyproject.toml`	Package configuration with all required dependencies
`src/envs/doom_env/tests/*.py`	Comprehensive test suite (4 test files, 65 total tests)
`examples/doom_example.py`	Example demonstrating Docker and local usage modes
`examples/doom_visualizer.py`	Interactive visualizer with keyboard controls (OpenCV)
`src/envs/doom_env/generate_gifs.py`	Utility script for generating scenario documentation GIFs
`src/envs/doom_env/README.md`	Extensive documentation (611 lines) with usage examples
`docs/environments.md`	Updated environment catalog with Doom card
`.github/workflows/docker-build.yml`	Added doom-env to CI/CD pipeline

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2025-12-09T23:00:28Z

src/envs/doom_env/models.py

+    game_variables: List[float] = None
+    available_actions: List[int] = None


Default values for fields with mutable types should use field(default_factory=...) instead of direct assignment. Using None as a default for List[float] and List[int] fields can lead to issues. These should be Optional[List[float]] or use field(default_factory=list).

Copilot · 2025-12-09T23:00:29Z

src/envs/doom_env/__init__.py

+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+"""Doom Env Environment - A simple test environment for HTTP server."""


The description says "Doom Env Environment - A simple test environment for HTTP server" but this is actually a full-featured ViZDoom integration for visual RL research, not a simple test environment. Update the description to accurately reflect its purpose.

Suggested change

"""Doom Env Environment - A simple test environment for HTTP server."""

"""Doom Env Environment - A full-featured ViZDoom integration for visual reinforcement learning (RL) research."""

Copilot · 2025-12-09T23:00:29Z

examples/doom_visualizer.py

+            # Random action for demonstration
+            # if result.observation.available_actions:
+            #     action_id = int(np.random.choice(result.observation.available_actions))
+


This comment appears to contain commented-out code.

Suggested change

# Random action for demonstration

# if result.observation.available_actions:

# action_id = int(np.random.choice(result.observation.available_actions))

Copilot · 2025-12-09T23:00:30Z

examples/doom_visualizer.py

+        action_id = 0  # Default: no action
+
+        if key == ord("q") or key == 27:  # Q or ESC
+            running = False


Variable running is not used.

Suggested change

running = False

Copilot · 2025-12-09T23:00:30Z

examples/doom_visualizer.py

+
+    result = env.reset()
+    plt.ion()
+    fig = plt.figure(figsize=(10, 7))


Variable fig is not used.

Suggested change

fig = plt.figure(figsize=(10, 7))

plt.figure(figsize=(10, 7))

Copilot · 2025-12-09T23:00:33Z

src/envs/doom_env/tests/test_doom_integration.py

+import subprocess
+import requests
+import numpy as np
+from pathlib import Path


Import of 'Path' is not used.

Copilot · 2025-12-09T23:00:33Z

src/envs/doom_env/tests/test_doom_environment.py

+"""
+
+import pytest
+import numpy as np


Import of 'np' is not used.

Copilot · 2025-12-09T23:00:33Z

src/envs/doom_env/tests/test_models.py

+Tests data model validation, serialization, and edge cases.
+"""
+
+import pytest


Import of 'pytest' is not used.

Copilot · 2025-12-09T23:00:34Z

src/envs/doom_env/client.py

+
+                    plt.close("all")
+                except ImportError:
+                    pass


'except' clause does nothing but pass and there is no explanatory comment.

Suggested change

pass

# Neither cv2 nor matplotlib is available for cleanup.

print(

"Warning: Could not clean up render windows because neither cv2 nor matplotlib is available. "

"Install with: pip install opencv-python or pip install matplotlib"

)

Copilot · 2025-12-09T23:00:34Z

src/envs/doom_env/server/doom_env_environment.py

+                    import matplotlib.pyplot as plt
+
+                    plt.close("all")
+                except ImportError:


'except' clause does nothing but pass and there is no explanatory comment.

Suggested change

except ImportError:

except ImportError:

# If matplotlib is not installed, we cannot close its windows, but this is non-critical during cleanup.

AlirezaShamsoshoara added 30 commits November 24, 2025 14:29

add the .dockerignore file

a2c6225

add the readme file

1bc2c00

add the init for the doom

791b46f

add the uv.lock

45e99e3

add the client code

0b754bb

add the model

9ff9988

add the openenv yaml file to doom env

5f86fd5

add the pyproject toml file for doom

602ea5a

add the DockerFile for Doom server

a198cf8

add the doom env environment to the server path

85b6dae

add the app for the server

142c921

add the init for the server - Doom

caf9358

add the example to the doom

1d4ce86

add vizdoom.ini to gitignore

d3c4a2b

give the opportunity to the user to be able to render the observation

74d9825

update the README for the new docker format

351bbaf

update the Dockerfile

c5ad51a

add the requirements for the server to install them in the docker image

55cdb2d

add the doom slayer drawing

7071952

a code to generate gifs to cover default scnearios

1b17ec4

add all gifs to the assets

e66f1de

add the gifs to top of the README file

e5c939e

update gif files and the project structure in README

3ca8ba9

add two more gifs for symmetric view on README

08a6eb1

move the example from the root of env to the example folder

92c96ab

fix relative path issue for importing

faf6bdc

update pyproject.toml to consider standard for uvicorn

d8b0fab

fix the issue of the _step_payload that it cannot serialize to json p…

5ae10c5

…roperly

support the visualization when running the docker

cd7292b

fix the issue of wrong visualization resolution for the docker run

3aa48bd

AlirezaShamsoshoara added 11 commits November 25, 2025 10:21

update the usage for the doom visualizer file

fc3a9d5

add doomenv to the docker build yaml

8a22dce

add the doom.md to the doc path

6fac432

update the lock and app to be compatible with openenv docker push and…

cb35e3d

… verifcation

fix the color issue in readme for openenv push feature

dec7e6c

update the dockerfile to fix the issue of relative path

03aaf74

add the environment card for the doom_env in the doc environments

e6eeb6f

add doom guy drawing

046e4b4

resize the image and align to the center via html

5a0cf44

align the project structure

6e0cf9c

update the README file to include the HF OpenEnv space for Doom-env

f45dd98

meta-cla bot added the CLA Signed This label is managed by the Meta Open Source bot. label Nov 26, 2025

AlirezaShamsoshoara added 7 commits November 30, 2025 17:11

start the unittest part for the doom - add init

235a3fd

add the conftest for the class types

5f36434

add the test models

87c571e

add the test doom client

da22ab4

add the test for doom environment

f131b62

add the test for doom integration with a server setup

f68e802

update the README for doom unittests

56ae7dd

init27 reviewed Dec 9, 2025

View reviewed changes

.github/workflows/docker-build.yml Show resolved Hide resolved

init27 requested a review from Copilot December 9, 2025 22:53

Copilot started reviewing on behalf of init27 December 9, 2025 22:54 View session

Copilot AI reviewed Dec 9, 2025

View reviewed changes

AlirezaShamsoshoara self-assigned this Dec 9, 2025

AlirezaShamsoshoara added the enhancement New feature or request label Dec 9, 2025

		game_variables: List[float] = None
		available_actions: List[int] = None

	"""Doom Env Environment - A simple test environment for HTTP server."""
	"""Doom Env Environment - A full-featured ViZDoom integration for visual reinforcement learning (RL) research."""

	# Random action for demonstration
	# if result.observation.available_actions:
	# action_id = int(np.random.choice(result.observation.available_actions))

	fig = plt.figure(figsize=(10, 7))
	plt.figure(figsize=(10, 7))

-                    pass
+                    # Neither cv2 nor matplotlib is available for cleanup.
+                    print(
+                        "Warning: Could not clean up render windows because neither cv2 nor matplotlib is available. "
+                        "Install with: pip install opencv-python or pip install matplotlib"
+                    )

	except ImportError:
	except ImportError:
	# If matplotlib is not installed, we cannot close its windows, but this is non-critical during cleanup.

[Enhancement] Add Doom-Env - Doom Slayer goes to OpenEnv School to learn RL #228

Are you sure you want to change the base?

[Enhancement] Add Doom-Env - Doom Slayer goes to OpenEnv School to learn RL #228

Conversation

AlirezaShamsoshoara commented Nov 26, 2025

Add Doom Environment with ViZDoom Integration

Summary

Overview

What's Included

Core Implementation

Docker Support

Documentation

Utilities

Assets

Key Features

1. Multiple Scenarios

2. Flexible Configuration

3. Action Spaces

4. Rendering Options

5. Docker Deployment

Technical Details

Architecture

Fixed Issues

Dependencies

File Structure

Usage Examples

Basic Usage

Docker Mode

With Rendering

Visual Examples

Testing

Documentation Updates

Validation

Related Issues

Future Work

Links

Breaking Changes

Checklist

Uh oh!

AlirezaShamsoshoara commented Nov 26, 2025

Uh oh!

AlirezaShamsoshoara commented Nov 26, 2025

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Dec 9, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 9, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 9, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 9, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 9, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 9, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 9, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 9, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 9, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 9, 2025

Choose a reason for hiding this comment

Uh oh!

Reviewers