Powered by the emerging large language models (LLMs), Autonomous GIS agents have the potential to accomplish spatial analyses and cartographic tasks. However, a research gap exists to support fully autonomous GIS agents: how to enable agents to discover and download the necessary data for geospatial analyses. This study proposes an autonomous GIS agent framework capable of retrieving required geospatial data by generating, executing, and debugging programs. The framework utilizes the LLM as the decision-maker, selects the appropriate data source (s) from a pre-defined source list, and fetches the data from the chosen source. Each data source has a handbook that records the metadata and technical details for data retrieval. The proposed framework is designed in a plug-and-play style to ensure flexibility and extensibility. Human users or autonomous data scrawlers can add new data sources by adding new handbooks. We developed a prototype agent based on the framework, released as a QGIS plugin (GeoData Retrieve Agent) and a Python program. Experiment results demonstrate its capability of retrieving data from various sources including OpenStreetMap, administrative boundaries and demographic data from the US Census Bureau, satellite basemaps from ESRI World Imagery, global digital elevation model (DEM), weather data from a commercial provider, the COVID-19 cases from the NYTimes GitHub. Our study is among the first attempts to develop an autonomous geospatial data retrieval agent.
Demonstration
- Video demonstrations on the data downloading examples can be found on our youtube channel.
We tested various data cases: by accepting data requests in natural language, most of the requests got correct data with an about 80% – 90% success rate. QGIS users can download the QGIS plugin (AutonomousGIS-GeodataRetrieveAgent) to download the data via natural language in a GIS environment. Note that for some data sources, you may need to apply API keys.
Figure below shows the QGIS plugin of the agent and the downloaded geospatial data of Nigeria, including cities (point), rivers (polyline), and state boundaries (polygon) from OpenStreetMap. The image basemap was downloaded from ESRI World Imagery using the plugin. Note that there are four individual data requests to retrieve the shown data (e.g., “Download the rivers in Nigeria.”).
The figure below shows the downloaded satellite images with the agent. Top map: the FAST Telescope in China (request: Download the FAST Telescope (Guizhou, China) satellite image at level 18.). Map 1: Yellowstone National Park (request: Download the Yellowstone National Park satellite image at level 10.); Map 2: a region defined by a bounding box (request: Download the satellite images of this region [south:27.82, west:86.73, north:28.17, east:87.13] at level 10.); Map 3: Qingdao City in China (request: Download the Qingdao, Shandong, China satellite image at level 10). Note that the red boundaries in Map 1 and Map 3 were downloaded using the plugin by two separate requests based on the place names: Download the boundary of Yellowstone National Park and Download the boundary of Qingdao, Shandong, China.
To learn more about the geospatial data downloading agent
- The source code of the LLM-Find agent (Python program with Jupyter Notebook) along with over 70 data request examples can be found at
https://github.com/gladcolor/LLM-Find. - The source code of the QGIS plugin can be found at: https://plugins.qgis.org/plugins/AutonomousGIS_GeodataRetrieverAgent/
- For more technical details, please refer to our preprint paper: Ning, Huan, Zhenlong Li, Temitope Akinboyewa, and M. Naser Lessani. 2024. “LLM-Find: An Autonomous GIS Agent Framework for Geospatial Data Retrieval.” arXiv. https://doi.org/10.48550/arXiv.2407.21024.
- The QGIS plugin can be downloaded from the official QGIS plugin repository at: https://plugins.qgis.org/plugins/AutonomousGIS_GeodataRetrieverAgent/
Note: LLM-Find is under active development, and the ideas presented in the paper may change due to the rapid development of AI. We hope LLM-Find can inspire the geospatial community to investigate autonomous GIS further.
