GEAR Lab Research on Multi-Agent Disaster Intelligence Accepted at ACM SIGSPATIAL 2026 🎉

GEAR Lab is pleased to share that our latest research paper has been accepted as a full Applications Track paper at the 34th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (SIGSPATIAL 2026) in Riverside, California. The work is co-authored by Yifan Yang, Wenjing Gong, Kaili Zhang, Lei Zou, Zhengzhong Tu, Hao Li, Zongrong Li, and Xinyue Ye.  

The paper, titled “RAPID: A Reproducible Multi-Agent Pipeline for Interpretable Disaster Damage Assessment from Satellite and Street-View Imagery,” introduces an autonomous multi-agent framework designed for zero-shot, interpretable post-disaster assessment. Addressing the challenges of domain shifts and heterogeneous geospatial data, RAPID coordinates four specialized agents—Disaster Perception (DPA), Image Restoration (IRA), Damage Recognition (DRA), and Disaster Reasoning (DReA)—to seamlessly integrate top-down remote sensing with ground-level street-view imagery.  

This work demonstrates how agentic AI and vision-language models can move disaster intelligence beyond black-box pattern recognition toward auditable, evidence-grounded spatial reasoning. Without requiring task-specific fine-tuning, the framework achieves robust multi-hazard classification, cross-view damage severity prediction, and fine-grained object recognition, while automatically generating location-specific, actionable disaster reports to support early-stage emergency response and recovery planning.  

The study highlights the potential of combining multimodal GeoAI, autonomous agent workflows, and cross-view geospatial observations to tackle critical disaster resilience challenges. As communities face increasingly frequent and severe climate hazards, frameworks like RAPID provide scalable, transparent decision-support tools for emergency managers, planners, and first responders.  

Congratulations to the research team on this contribution to autonomous GeoAI and disaster informatics!  

Citation

Yang, Y., Gong, W., Zhang, K., Zou, L., Tu, Z., Li, H., Li, Z., & Ye, X. (2026). RAPID: A Reproducible Multi-Agent Pipeline for Interpretable Disaster Damage Assessment from Satellite and Street-View Imagery. In Proceedings of the 34th ACM International Conference on Advances in Geographic Information Systems (SIGSPATIAL '26), November 3–6, 2026, Riverside, CA, USA. ACM. https://doi.org/10.1145/3841645.3843346