Path planning algorithm for mobile robots in high-speed railway waiting halls under boarding-gate congestion

Purpose This paper aims to address the path-planning problem of a single mobile robot performing transport tasks in a high-speed railway waiting hall under boarding-gate congestion. Unlike general indoor navigation, this scenario involves schedule-driven localized congestion, semantic differences between target and non-target gates and a strict pre-departure deadline. The study seeks to develop a path-planning method that can improve execution efficiency, reduce unnecessary exposure to irrelevant congestion areas and better satisfy the operational requirements of in-station transport tasks. Design/methodology/approach A scenario-adapted path-planning method based on time-space A* is proposed. First, a gate-centered time-varying congestion model is constructed by integrating train departure schedules, ticket-checking time windows and spatial decay. Then, a task-oriented semantic mapping strategy is introduced to distinguish necessary approach to the target gate from avoidable exposure to non-target gates. Finally, a search framework combining time-varying traversal cost, semantic exposure penalty and deadline-aware pruning is developed for single-robot transport tasks in waiting halls. Findings Experimental results in normal, peak and dense scenarios with 30 random seeds show that the proposed method consistently achieves 100% on-time completion and feasible-path rates. Compared with static shortest-path methods, it reduces unnecessary exposure and improves execution efficiency. Compared with RRT*, it shows better feasibility and stability under deadline constraints. Compared with D* Lite, it achieves lower SimTime, much lower exposure and lower planning cost. The findings confirm the effectiveness of task-oriented time-space A* adaptation for predictable localized congestion in waiting halls. Research limitations/implications This study is limited to a single-robot, single-task setting in a simulated high-speed railway waiting-hall environment. Multi-robot coordination, task allocation, elevator resource competition and real-time sensing feedback are beyond the current scope. The congestion model is mainly timetable-driven and assumes known station layout, gate positions and train-to-gate assignments. These limitations suggest that future research should incorporate online perception, dynamic updates and more realistic deployment conditions to further strengthen the applicability of the proposed method. Practical implications The proposed method provides a practical planning framework for intelligent in-station transport tasks such as parcel transfer, small-item delivery and other short-distance logistics operations in high-speed railway waiting halls. By reducing unnecessary traversal through irrelevant congestion hotspots and improving deadline compliance, the method can enhance operational efficiency and planning quality in congestion-sensitive station environments. It also offers a useful reference for integrating timetable information and localized congestion priors into practical robotic navigation systems deployed in smart-station scenarios. Social implications This study contributes to the development of safer, more efficient and more intelligent railway-station service systems. Improved robot navigation under boarding-gate congestion can help reduce operational disruption, support more reliable in-station logistics and promote the coordinated use of robotic systems in public transport hubs. In the longer term, such methods may support smarter passenger-service infrastructure, improve service quality in busy railway environments and facilitate the digital transformation of high-speed railway stations toward more resilient and intelligent operation. Originality/value The originality of this study lies in adapting time-space A* to the specific operational characteristics of high-speed railway waiting halls. Unlike generic indoor navigation methods, the proposed approach explicitly models gate-centered timetable-driven congestion, distinguishes the semantic roles of target and non-target gates and integrates deadline-aware pruning into the search process. The study therefore provides a task-oriented path-planning framework tailored to waiting-hall transport tasks, offering both methodological value for congestion-aware robot navigation and practical value for intelligent railway-station applications.