This is the 1st place solution for 3D Open World FPS Game AI Chanllenge(2022 cog&inspir.AI)
Admittedly, our code currently is far from perfection. It still have large potentials to be completed and developed.
-
agent_for_track1a.py
andagent_for_track1b_and_2.py
provide rule-based solution for the track1-1, track1-2 and track2 respectively. -
agent_RLsimple.py
gives Reinforcement Learning solution, which can be trained bytrain_ppo.py
.
Environment used in this task and the evaluation code can be found at https://github.com/inspirai/wilderness-scavenger