Work

Dynamic Multi-Robot Task Allocation in Multi-Floor Hospitals: An Urgency-Aware Auction Framework

Year: 2026
Type: preprint
Abstract: Multi-robot hospital delivery involves coordinating the transport of medical supplies across different floors while accommodating dynamically arriving requests and urgent delivery demands. Inter-floor elevator transfers and the need to serve multiple destinations make task allocation dependent not only on robot assignments but also on the composition and execution order of delivery queues. To address these challenges, this paper proposes a Greedy Coalition Auction Algorithm (GCAA) for dynamic task allocation in multi-floor hospitals. The algorithm constructs capacity-limited task queues through marginal-utility-based bidding, accounting for the travel costs associated with serving multiple destinations in a single delivery trip. The scheduling framework combines batch queue construction with online task adjustment. Newly arriving ordinary tasks are considered in subsequent allocation opportunities, while urgent requests trigger a preemption-and-rollback procedure that releases affected unfinished assignments for later scheduling. Experiments conducted in a multi-floor hospital simulation environment evaluate GCAA against adapted task-allocation methods and examine its performance under different workloads and hospital layouts. The results indicate that batch queue construction improves overall delivery timeliness, while urgent-task preemption supports timely responses to emergency requests at the expense of some ordinary-task performance. GCAA achieves favorable results relative to the adapted CBBA and DSTA methods on several measures, although the adapted ADA method achieves better overall service and travel performance. These findings illustrate the trade-offs between routine delivery efficiency and urgent-task response in dynamic multi-robot hospital scheduling.
Language: English
Open Access status: green
OpenAlex ID: W7236497544
Cites: 37
Cited by:
Related to: 10