Fast and Precise: Adjusting Planning Horizon with Adaptive Subgoal Search
Micha{\l} Zawalski, Micha{\l} Tyrolski, Konrad Czechowski, Tomasz, Odrzyg\'o\'zd\'z, Damian Stachura, Piotr Pi\k{e}kos, Yuhuai Wu, {\L}ukasz, Kuci\'nski, Piotr Mi{\l}o\'s

TL;DR
AdaSubS is an adaptive planning method that dynamically adjusts subgoal distances to efficiently solve complex reasoning tasks by balancing long-term planning and short-term control.
Contribution
The paper introduces AdaSubS, a novel adaptive subgoal search algorithm that filters unreachable subgoals to improve planning efficiency in complex problems.
Findings
AdaSubS outperforms hierarchical planning algorithms on Sokoban, Rubik's Cube, and INT.
Adaptive adjustment of planning horizon enhances scalability and efficiency.
Filtering unreachable subgoals accelerates the planning process.
Abstract
Complex reasoning problems contain states that vary in the computational cost required to determine a good action plan. Taking advantage of this property, we propose Adaptive Subgoal Search (AdaSubS), a search method that adaptively adjusts the planning horizon. To this end, AdaSubS generates diverse sets of subgoals at different distances. A verification mechanism is employed to filter out unreachable subgoals swiftly, allowing to focus on feasible further subgoals. In this way, AdaSubS benefits from the efficiency of planning with longer subgoals and the fine control with the shorter ones, and thus scales well to difficult planning problems. We show that AdaSubS significantly surpasses hierarchical planning algorithms on three complex reasoning tasks: Sokoban, the Rubik's Cube, and inequality proving benchmark INT.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
Taxonomy
TopicsAI-based Problem Solving and Planning · Logic, Reasoning, and Knowledge · Artificial Intelligence in Games
