2025 CORL CoRL 2025

$Door(s)$: Junction State Estimation for Efficient Exploration in Reinforcement Learning