iA∗: Imperative Learning-based A* Search for Pathfinding
Core Concepts
iA∗ proposes a novel pathfinding framework combining imperative learning and A* search to improve efficiency and reduce search area.
Abstract
I. Introduction
Pathfinding is crucial for robot navigation.
Classic methods like A* face scalability issues.
Data-driven approaches lack theoretical guarantees.
II. Methodology
iA∗ combines imperative learning with A* search.
Bilevel optimization process enhances efficiency.
III. Experiments
iA∗ surpasses classical and data-driven methods in efficiency.
Demonstrates superior robustness across tasks and environments.
iA$^*$
Stats
"Our comprehensive experiments demonstrate that iA∗ surpasses both classical and data-driven methods in pathfinding efficiency."
"The experimental results demonstrate that iA∗ outperforms both classical and learning-based pathfinding methods."
Quotes
"To improve search efficiency and add interpretability, we propose an IL-based A* search framework (iA*) that includes a bilevel optimization process."
"Experimental results demonstrate that the proposed iA* outperforms both classical and learning-based pathfinding methods."