Research on Adaptive Navigation System for Hospital Logistics AMR based on Active Trajectory Prediction

Abstract

In-hospital logistics is a core link in medical logistics support. Traditional manual delivery has problems such as low operational efficiency, difficult material traceability, and loopholes in hospital infection control. Existing autonomous mobile robots (AMRs) in hospital scenarios with dense pedestrian flow and complex spatial environments have bottlenecks such as insufficient navigation accuracy, weak dynamic obstacle avoidance, and poor spatial adaptability, making it difficult to meet actual needs. Therefore, this paper develops an intelligent AMR logistics system for hospital scenarios: integrating Simultaneous Localization and Mapping (SLAM) navigation technology, fusing YOLACT pedestrian detection and Interacting Multiple Model (IMM) tracking algorithms to achieve dynamic pedestrian perception and trajectory prediction, and combining the Time Elastic Band (TEB) planning algorithm to complete dynamic obstacle avoidance and path optimization, which improves the passage efficiency and navigation robustness in high pedestrian density scenarios; the robot body adopts a fusion design of an omnidirectional chassis and a Stewart parallel platform, taking into account the maneuverability in narrow spaces, realizing pose compensation under extreme working conditions, and reducing transportation losses; a digital twin management platform is built to cover the full process of intelligent scheduling, real-time positioning, and material traceability, and an integrated closed-loop management system is constructed. This system solves the pain points of in-hospital logistics and the technical defects of existing robots, improves the delivery efficiency and hospital infection control level, and provides a feasible solution and practical reference for the intelligent upgrade of hospital logistics.