Background, motivation, and the evolution of autonomous mobile robot systems

Abstract

Autonomous mobile robots (AMRs) have emerged as a central technology in modern robotics, integrating perception, localization, mapping, planning, and control to enable robust autonomous navigation in complex and dynamic environments. Recent advances in multi-modal sensing, deep-learning-based perception, optimization-driven and learning-enhanced planning, and hierarchical control architectures have significantly expanded the capabilities and deployment scope of AMRs across industrial logistics, manufacturing, healthcare, agriculture, service robotics, and urban mobility. This survey provides a comprehensive review of the algorithmic foundations, system architectures, and practical applications that define contemporary AMR research. We examine the evolution of perception pipelines from geometric methods to multimodal and semantic understanding, analyze major developments in SLAM, including graph-based optimization and learning-augmented pipelines, and discuss global and local planning frameworks encompassing heuristic search, sampling-based algorithms, trajectory optimization, and reinforcement learning. We further investigate control techniques ranging from classical nonlinear control and MPC to safety-critical and hybrid learned controllers. In addition, we highlight integration challenges, real-world deployment issues, and emerging research directions such as lifelong autonomy, human-aware navigation, multi-robot collaboration, and the rise of foundation models for robotics. This survey aims to provide a unified perspective that can guide researchers and practitioners in advancing next-generation AMR systems capable of reliable long-term operation in real-world settings.