Purpose The purpose of this study is to detail the design and development of a robust and practical perception system for autonomous material handling robots (AMHRs) operating within industrial stockyards. This system aims to support simultaneous localization and mapping (SLAM) while generating large-scale spatial cognition, ensuring accurate, low-latency, and scalable operations in demanding industrial environments. Design/methodology/approach The proposed perception system integrates multimodal perception sensors, efficient algorithms and commercial hardware devices to o provide SLAM-based large-scale spatial cognition for distributed AMHRs. The system’s design emphasizes practicality, efficiency and readiness for real-world deployment, ensuring it meets the stringent requirements of accuracy, latency and scalability. Findings Experiments conducted in a real industrial stockyard environment demonstrate the practicality and robustness of the perception system. The system exhibits high performance in state estimation, stockpile modeling accuracy and motion spatial cognition, confirming its effectiveness for AMHR operations. Practical implications The developed system was practically used at Tianjin Port, demonstrating the potential for widespread industrial application, offering a scalable and efficient solution for AMHR operations. Its integration into diverse industrial settings can lead to significant improvements in material handling processes, contributing to enhanced productivity and operational efficiency. Originality/value This work presents an innovative perception system that combines advanced SLAM-based spatial cognition akin to that of the brain with practical deployment considerations. The system’s design and implementation address the specific challenges of AMHRs in industrial environments, providing a novel solution for enhancing the operational efficiency and adaptability of autonomous robots in stockyards and similar settings.
Read full abstract