FloodPulse

Name: FloodPulse
Author: Mitchell-Odili

0•

Orchestrates multimodal AI agents in an offline-first simulation to enhance urban flood resilience, particularly within Nairobi's Mbagathi Basin.

FloodPulse addresses the critical 'data darkness' gap during urban floods by providing an offline-first, multimodal resilience assistant. It leverages edge AI to enable 'Digital Guardian' protocols, allowing devices to 'see' and 'reason' about flood risks without relying on active internet. The tool validates these protocols through a modular agentic simulation, modeling the Mbagathi River basin with parametric personas (Commuter, Responder, Strategist) and advanced geospatial AI to identify safe ridges and critical flood zones, ensuring local truth is accessible when traditional navigation fails.

主な機能

01Zero-shot vision analysis using Gemma 4 for spatial reasoning and flood zone identification

020 GitHub stars

03Offline-first multimodal resilience for critical infrastructure failure

04Parametric Persona Engine for consistent agent identities (Commuter, Responder, Strategist)

05Multi-sensor fusion (spectral analysis, NASA SRTM, SAR) to detect hidden river paths and standing water

06Modular Agentic Simulation for validating multi-agent consensus in high-risk scenarios

ユースケース

01Validating 'Digital Guardian' protocols for urban flood resilience using multimodal LLMs

02Simulating multi-agent rescue and evacuation scenarios in flood-prone urban environments

03Providing offline, spatial-aware navigation assistance to find safe routes during flash floods