From Classical Search to Intelligent Agents: An Investigation of Maze Solving Algorithms
ITM Web of Conferences · 2026 · 93 · 02012
DOI: 10.1051/itmconf/20269302012 ↗Kurz gefasst
Fachartikel in der Zeitschrift ITM Web of Conferences, online erschienen am 13.11.2026. Verfasst von Jingran Zhang. Die Arbeit nennt 11 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Finding paths through mazes is a classic AI problem that tests how algorithms handle limited information and resource constraints. This survey reviews the evolution of maze-solving methods from classical search (BFS, DFS, Dijkstra, A*) to reinforcement learning (Q-Learning, DQN) and multi-agent systems. This paper examines how each approach works, what resources it needs, and where it performs best. Classical algorithms give optimal paths in known, static mazes but fail when the environment changes or scales up. Deep RL offers adaptability by learning from experience without a map. Multi-agent systems add parallel exploration and robustness, yet they bring new overhead. This paper identifies four persistent challenges: interpretability, generalization, training cost, and coordination complexity. This paper then discusses promising directions: hybrid models, explainable AI, transfer learning, and edge-deployment techniques. Together, these directions suggest that future maze-solving research should balance theoretical optimality, adaptive learning, computational efficiency, and practical deployment in uncertain real-world navigation scenarios.
Zitieren
APA
Zhang, J. (2026). From Classical Search to Intelligent Agents: An Investigation of Maze Solving Algorithms. ITM Web of Conferences, 93, 02012. https://doi.org/10.1051/itmconf/20269302012
Vancouver
Zhang J. From Classical Search to Intelligent Agents: An Investigation of Maze Solving Algorithms. ITM Web of Conferences. 2026;93:02012. doi:10.1051/itmconf/20269302012
BibTeX
@article{Zhang2026,
author = {Zhang, Jingran},
title = {From Classical Search to Intelligent Agents: An Investigation of Maze Solving Algorithms},
journal = {ITM Web of Conferences},
year = {2026},
volume = {93},
pages = {02012},
doi = {10.1051/itmconf/20269302012}
}
EUVIMED