An AI agent trained to navigate obstacle towers
Stairway to Heaven is an AI agent that seeks to quickly and efficiently navigate floors of increasing difficulty as it ascends an obstacle tower. Our primary objective is to train an agent to navigate as many floors as possible of a static environment. Our secondary objectives are to apply this learning to random environments, and should time permit, analyze the qualities of solved and attempted puzzles to understand the correlation between the abilities of the agent and the limitations of reinforcement learning.