人工智慧賦能的自主移動機器人導航：感知、 規劃與基於學習的控制研究調查

隨著自主移動機器人 (AMR) 在智慧製造的快速普及，自主導航已成為AMR 的核心能力。本文探 討人工智慧導入自主移動機器人導航系統之發展趨勢，聚焦於感知、規劃與學習式控制三大核心模組。 傳統 AMR 導航多仰賴規則式與模型式方法，於動態與複雜場域中會有侷限。近年深度學習與強化學習技 術被引入導航本體架構，使機器人具備語意感知、學習式路徑規劃與端對端運動控制能力。本文透過系 統性文獻回顧，分析 AI 如何重塑導航流程，並整理多模態感測融合、神經代價地圖與強化學習控制等關 鍵技術，最後探討整合式 AI 導航架構之挑戰與未來發展方向。

With the rapid popularization of autonomous mobile robots (AMRs) in smart manufacturing, autonomous navigation has become a core capability of AMRs. This article explores the development trend of integrating artificial intelligence into autonomous mobile robot navigation systems, focusing on three core modules: perception, planning, and learning-based control. Traditional AMR navigation relies heavily on rule-based and model- based methods, which have limitations in dynamic and complex environments. In recent years, deep learning and reinforcement learning technologies have been introduced into navigation ontology architectures, enabling robots to possess semantic perception, learning-based path planning, and end-to-end motion control capabilities. This article analyzes how AI is reshaping the navigation process through a systematic literature review, summarizes key technologies such as multimodal sensing fusion, neural cost mapping, and reinforcement learning control, and finally discusses the challenges and future development directions of integrated AI navigation architectures.