OPEN: Lightweight Map-Based Semantic Navigation for GPS-Free Last-Mile Delivery

The growing demand for efficient last-mile delivery highlights the need for autonomous robots to improve operational efficiency and reduce costs. Traditional navigation methods rely on high-precision maps, which are expensive to produce and maintain, while learning-based approaches often struggle to adapt to diverse real-world environments. To address these challenges, this paper presents OpenStreetMap-enhanced oPen-air sEmantic Navigation (OPEN), a system that combines foundation models with classic navigation algorithms to enable scalable outdoor navigation. By leveraging off-the-shelf OpenStreetMap (OSM), OPEN eliminates the need for extensive pre-mapping and provides a lightweight, readily available map representation. The system uses Large Language Models (LLMs) to interpret delivery instructions and Vision Language Models (VLMs) for global localization without relying on GPS, real-time map updates, and entrance recognition, ensuring robust navigation in complex environments. To further enhance adaptability, OPEN incorporates a local replanning method that dynamically adjusts waypoints in response to environmental changes and OSM inaccuracies. Since existing benchmarks do not adequately reflect the challenges of last-mile delivery, this work introduces a new benchmark designed for residential navigation. Experiments conducted in both simulated and real-world settings demonstrate that OPEN improves navigation accuracy, efficiency, and reliability, outperforming existing semantic navigation methods. To facilitate further research, the code and benchmark are publicly available at https://github.com/EI-Nav/light-map-navigation.git Note to Practitioners—Autonomous robot navigation for last-mile delivery faces significant challenges, including high mapping costs, environmental variability, and unreliable or unavailable GPS signals in urban areas, all of which hinder large-scale deployment. Traditional high-precision maps are expensive to create and maintain, while learning-based approaches often struggle to generalize to new and dynamic environments. Additionally, traditional navigation systems lack the ability to comprehend and execute tasks based on natural language instructions, limiting their flexibility and interactivity. To address these challenges, this work introduces an innovative framework that eliminates the need for costly pre-mapping by leveraging OSM data and foundation models for efficient navigation. The system utilizes LLMs to interpret delivery tasks expressed in natural language, offering a more intuitive and interactive user experience, while VLMs enable GPS-free localization, real-time map updates, and entrance recognition, enhancing adaptability in complex environments. Additionally, a local replanning mechanism dynamically adjusts waypoints when OSM data proves inaccurate, ensuring greater safety and reliability. In practical applications, OPEN provides a low-cost, highly adaptable solution that allows robots to operate in diverse environments without extensive pre-mapping efforts, while its ability to process natural language instructions significantly enhances usability. However, its effectiveness depends on the completeness and accuracy of OSM data, and integrating higher-quality maps from providers like Google or Baidu could further improve navigation performance. Beyond last-mile delivery, this framework holds promising potential for applications in drone logistics, warehouse automation, and intelligent transportation systems, paving the way for scalable and cost-effective autonomous mobility solutions.