From sensor signal to city action

See the flood
before the street does.

FloodLens fuses rainfall, water-level, flow, and sewer-obstruction signals into an explainable neighborhood forecast—then tells communities and responders what to do next.

Open command center ↘Built from an award-winning system →

4signals fused

93%decision confidence

< 2sedge inference

Jakarta digital twin

JAKARTA BARATJAKARTA UTARAJAKARTA TIMUR

LOWSEVERE

ORANGE protocolKampung Melayu4 min response window

ULTRASONIC NODE 04•WATER LEVEL 78%•FLOW 41 L/S•RISING RAPIDLY•CONFIDENCE 93%

Live decision workspace

One city. Four signals.
One clear decision.

Run a transparent sensor-fusion scenario. Every score includes its drivers, uncertainty, escalation protocol, and recommended action.

Neighborhood risk6-hour horizon

93% confidence

74/100

ORANGE PROTOCOL

Heightened flood risk

Conditions are deteriorating. Clear drainage constraints and prepare community alerts while monitoring the next sensor window.

Estimated action window4 minutes

River pressure68

78% ultrasonic level

Rain load59

118 mm accumulated

Drainage stress41

34% obstruction

Recommended response chainOrdered by urgency

Warn residents

Issue localized SMS alert in Bahasa Indonesia and English.

✓

Clear the flow

Dispatch drainage crew to Ciliwung catchment.

✓

Prepare evacuation

Open the nearest verified shelter and accessible route.

✓

Trigger actuator

Arm pump or mechanical response after human approval.

✓

Sensor-to-action network

Built for the places
weather apps cannot see.

Flooding begins at street and sewer level. FloodLens closes that last-mile visibility gap with inexpensive edge nodes and accountable decisions.

Node JL-04 · CiliwungLIVE

Water-level signal

Raw sensorSmoothed trendDanger threshold

HC
SR04

Edge node anatomy

Low-cost sensing.
High-context decisions.

Ultrasonic water-level measurement
Local smoothing and anomaly filtering
Offline-safe yellow, orange, red protocol
Cloud telemetry for citywide coordination

01Sense

Ultrasonic and analog nodes sample sewer conditions.

→

02Filter

Moving average and anomaly checks remove sensor noise.

→

03Infer

Signal fusion estimates risk and decision confidence.

→

04Act

Alerts, crews, pumps, and evacuation follow explicit rules.

A three-year engineering journey

Not a weekend idea.
A system that kept evolving.

FloodLens continues Shannon Aurelia Widjaja’s work on flood resilience—from an award-winning high-school smart-city proposal to a physical embedded prototype, and now an explainable decision layer for Jakarta.

2023

I²Tech Olympiad · IoT & Data Analytics

Smart Sewer concept wins

Proposed ultrasonic sewer sensing, Arduino-to-cloud telemetry, municipal dashboards, SMS alerts, pump actuation, and a three-level response protocol for Jakarta.

2025

Universitas Indonesia · Embedded Systems

The concept becomes hardware

Built and tested a water-level detector with LEDs, buzzer, LCD, servo response, time-series logging, smoothing, peak detection, and anomaly analysis.

2026

FloodLens · AI for Earth

Hardware becomes intelligence

Fuses environmental and infrastructure signals, quantifies uncertainty, explains every prediction, and turns risk into an accountable response chain.

The measurable thesis

Earlier warning is only useful when it creates earlier action.

4→1sensor streams become one actionable risk state

3offline-safe escalation levels inherited from Smart Sewer

100%of predictions expose drivers and confidence

Prototype transparency

This public demonstration uses configurable scenarios and a deterministic, interpretable sensor-fusion model. It does not claim to replace official BMKG, BPBD, or field measurements. The next validation stage pairs deployed nodes with labeled local flood events.