This paper investigates fundamental concepts related to detecting and mitigating partial observability by measuring misalignment between value function estimates. The paper was presented at the “Finding the Frame” workshop at RLC 2024 and the “Foundations of Reinforcement Learning and Control” workshop at ICML 2024.

The work was jointly led by Cameron Allen, Aaron T. Kirtland, and Ruo Yu Tao, in collaboration with Sam Lobel, Daniel Scott, Nicholas Petrocelli, Omer Gottesman, Ronald Parr, Michael L. Littman, and George Konidaris.

Check out the videos at the project page, or read the full paper here.