Mario_RL is a course project developing a reinforcement learning (RL) agent for Super Mario Bros. It focuses on boosting agent performance via observation/action/reward space feature engineering (not advanced RL tweaks), using the DI-adventure baseline.
review DQN-related papers, use gym-super-mario-bros v0, run/evaluate baseline (PyTorch), test feature processing, analyze performance. Deliverables: report, A0 poster, code package. It requires online RL learning without pre-trained models/imitation data.