Artificial intelligence-driven optimization and reinforcement learning for intelligent mobile robot operations in warehouse automation: A review

Warehouse automation increasingly relies on mobile robots capable of perception, navigation, manipulation, and decision-making in dynamic environments. This paper reviews recent AI techniques—especially deep learning-based perception, reinforcement learning for navigation and task allocation, and optimization methods for multi-robot coordination—and explains how these technologies directly enhance the efficiency, accuracy, and reliability of robotic mobile fulfillment systems (RMFS). We propose an integrated AI pipeline combining object detection, grasp planning, motion planning, and fleet-level scheduling, and we analyse how state-of-the-art methods address core warehouse challenges such as cluttered bins, dynamic obstacles, routing conflicts, and energy constraints. By connecting AI techniques to concrete robotic capabilities, this review identifies essential technological drivers for next-generation warehouse mobile robots and highlights open research gaps that require further investigation. Unlike existing surveys that examine perception, navigation, manipulation, or scheduling in isolation, this review integrates all key components of warehouse mobile robot intelligence into a unified pipeline. By connecting perception → navigation → grasping → task allocation → RMFS → energy management, the paper provides a warehouse-specific, end-to-end perspective missing from earlier works. This holistic analysis highlights cross-dependencies, performance trade-offs, and research gaps across the entire robotic workflow.