Development of an Indoor Delivery Mobile Robot for a Multi-Floor Environment

Abstract

In recent years, the demand for delivery services has increased by applying robot technology in various fields such as food services, logistics, hospitals, and hotel business. However, it is still challenging to perform autonomous delivery in multi-floor buildings. Particularly for wheeled robots, the use of elevators is essential for indoor last-mile delivery service in buildings. To tackle the problem, We have developed an indoor delivery mobile robot and present its architecture designed for multi-floor environments. The architecture consists of five modules: map management for utilizing an integrated navigation map, localization, path planning, perception, and task planning. The integrated navigation map is generated by combining multi-floor point cloud maps and topological maps based on node graphs for effective localization and path planning. Additionally, the proposed 3D route planning allows inter-floor movement. A feasible path for boarding the elevator can be generated through the perception module, and delivery services to multiple destinations can be repeatedly performed through task planning. Our architecture's effectiveness is demonstrated through a month-long field test in an ordinary building during regular business hours. This study's contributions include a novel architecture for autonomous delivery without human intervention, an integrated map for efficient indoor navigation, and the proven robustness of the system in real-world scenarios.