IoT Pothole Detection using Ultrasonic Sensor

Potholes damage vehicles and cause accidents, yet road-condition surveys in most cities are still manual, infrequent and subjective — an inspector drives a stretch and notes defects by eye. A low-cost sensing unit on ordinary vehicles could make defect reporting continuous and location-accurate: measure the road profile, flag sudden dips and bumps, attach GPS coordinates, and stream the events to a map the municipality can act on. This project builds the sensing-and-telemetry half of that vision at student scale: an ultrasonic sensor profiles the road surface, an accelerometer confirms the jolt, and an ESP32 publishes geo-tagged events to a live dashboard. The scope is honest — a demonstrator on a model vehicle, not a certified road-survey instrument — but the pipeline (sensing, fusion, geo-tagging, cloud map) is the real architecture such systems use.

1. The HC-SR04 continuously measures the distance from the vehicle underside to the road surface.
2. A rolling baseline of the road distance is maintained; deviations beyond the configured depth threshold start a candidate event.
3. The MPU6050's vertical acceleration is checked for a matching jolt spike within the same window to confirm the event.
4. The ESP32 reads the NEO-6M GPS for latitude and longitude and packages the event as JSON.
5. The event is published over Wi-Fi/MQTT to the broker and appended to the microSD backup log.
6. The dashboard subscribes to the event topic and drops a marker on the live map with depth, jolt and timestamp.

Tech stack:

• ESP32-WROOM-32, firmware in Arduino IDE (C++)
• HC-SR04 ultrasonic sensor (2–400 cm, ~±3 mm typical)
• MPU6050 accelerometer (jolt cross-check)
• NEO-6M GPS module (UART, NMEA)
• MQTT telemetry (Mosquitto broker)
• Node-RED/Blynk live map dashboard
• microSD event backup log
• 5 V vehicle power (buck converter from 12 V)

• [Ultrasonic road profiling] A downward-facing HC-SR04 (2–400 cm range, ~±3 mm typical accuracy) measures road distance; sudden increases flag potholes, sudden decreases flag bumps.
• [Accelerometer cross-check] The MPU6050's vertical-axis spike confirms the jolt, filtering out ultrasonic noise and momentary echoes.
• [GPS geo-tagging] A NEO-6M module stamps latitude and longitude on every confirmed event.
• [MQTT telemetry and live map] The ESP32 publishes events to an MQTT broker; a Node-RED/Blynk dashboard plots them as map markers in real time.
• [On-device event log] A microSD card backs up every event with timestamp and coordinates when Wi-Fi is unavailable.
• [Configurable thresholds] Pothole depth threshold and jolt g-threshold are set via a configuration file and dashboard controls — no firmware change needed.
• [Demo vehicle mount] An adjustable bracket fits the unit to a model car, powered from 5 V USB or a 12 V buck converter.

• Assembled sensing unit (ESP32, sensors, GPS, mount)
• ESP32 firmware source code with comments
• MQTT + dashboard setup guide
• Wiring diagram and component list
• Threshold-tuning procedure document
• Project abstract PDF
• Project report (background, sensor fusion logic, methodology, demo results)
• PPT presentation and viva Q&A document

• This is a demo-scale build on a model vehicle; validation at real traffic speeds on public roads is outside the project's scope.
• The HC-SR04's ~15° beam and speed-dependent echo quality limit it; wet or soft surfaces absorb sound and weaken returns.
• Rough-but-healthy roads can trigger false positives — thresholds need tuning for each surface, documented in the tuning guide.
• GPS needs open sky; urban canyons and flyovers degrade fix quality.
• Live telemetry needs a Wi-Fi hotspot in the vehicle; without it, events accumulate only on the SD card (GSM uplink is an optional add-on).