Planning Horizon
Planning horizon, the timeframe considered when making decisions, is a crucial parameter in various planning and reinforcement learning problems, impacting both computational efficiency and solution quality. Current research focuses on optimizing planning horizon dynamically, adapting it to problem complexity and uncertainty, often employing techniques like hierarchical planning, receding horizon control, and adaptive subgoal search within model architectures such as deep Q-networks and Monte Carlo tree search. These advancements aim to improve the efficiency and robustness of planning algorithms across diverse applications, from robotics and autonomous driving to online resource allocation and conversational AI systems.
18papers
Papers
March 26, 2025
September 26, 2024
August 13, 2024
October 23, 2023
December 30, 2022