What it does
This smart system monitors water levels and flow using sensors and AI, predicts floods, and sends real-time alerts via a mobile app. It aims to solve delayed and inaccurate flood warnings in vulnerable Malaysian regions.
Your inspiration
The inspiration came from Malaysia’s recurring and devastating floods, especially during the monsoon season, which displace thousands of people and cause massive damage annually. The December 2024 flood emphasised the urgent need for a better early warning system. Existing systems often lack real-time prediction, broad coverage, and user accessibility. Witnessing these gaps firsthand and learning from Jabatan Pengairan dan Saliran (JPS) sparked our motivation to create an accessible, AI-powered, sustainable, and scalable flood monitoring and alert system for both rural and urban communities.
How it works
The system integrates ultrasonic sensors and a camera with AI-based Support Vector Machine (SVM) algorithms to monitor water level and flow velocity. These sensors feed real-time data to a Raspberry Pi, which processes and transmits it via MQTT to a mobile app. The AI analyses historical and current data to predict floods. A solar panel powers the system, ensuring sustainable and uninterrupted operation even during outages. The app provides users with real-time flood risk percentages, location tracking, and alerts — ranging from early warnings (two hours ahead) to imminent flood notifications. By combining hardware and intelligent software, the system ensures accurate, timely, and area-specific flood alerts to enhance preparedness.
Design process
The project began with research on Malaysia’s flood issues and shortcomings in the current alert systems. We conducted a market survey targeting tech-savvy users in flood-prone areas to identify needs. From their feedback, we developed key design requirements such as real-time alerts, predictive AI, and GIS integration. Next, we explored existing patents and identified innovation gaps. We then brainstormed and shortlisted components based on criteria like accuracy, cost, power consumption, and scalability. The chosen design used an ultrasonic sensor and a camera with SVM for water monitoring, solar panels for power, and MQTT for communication. The mechanical structure was fabricated based on CAD designs, and all components were integrated and tested. We created a mobile app interface using IoT MQTT Panel to visualise real-time data. The system underwent simulation, validation, and refinements, especially in flood risk classification accuracy. Prototypes were tested under simulated water flow conditions, and alerts were triggered at various thresholds. Results showed reliable flood prediction and real-time updates.
How it is different
Unlike current flood systems that rely on manual updates or simple sensors, this design offers an integrated, AI-powered solution with real-time alerts and predictive analytics. It combines visual detection (via camera and SVM), ultrasonic measurement, and fuzzy logic to assess flood risk. Unlike systems limited to urban infrastructure (e.g., SMART Tunnel), this solution is solar-powered and deployable in rural areas with no power grid. It surpasses Public InfoBanjir and others by offering multi-area tracking, custom notifications, and real-time risk percentages through a user-friendly mobile app. The dual-sensor setup improves reliability in varied conditions, and the modular design ensures easy scalability. By automating data collection, prediction, and alerts, the system significantly improves early warning precision and accessibility.
Future plans
We plan to refine the AI model with a larger dataset for better accuracy and expand the system’s compatibility with multiple mobile platforms (iOS and Android). More sensors will be added for broader environmental monitoring. Collaborations with local authorities and JPS are envisioned to deploy the system in actual flood-prone areas. We aim to commercialise a plug-and-play version and integrate route-planning features for safe evacuation. Future enhancements also include integrating weather data and improving the user interface for better public engagement and disaster response.
Share this page on