Traffic Signal Preemption System for Emergency Vehicles using ESP32

Emergency vehicles lose critical minutes waiting at red lights, and real-world preemption systems (the kind that hold green lights for ambulances) cost far more than a student lab can justify and demand certified hardware. This project distills the core idea into a bench-top demonstrator: an ESP32 board, an electret microphone module, and a model traffic light showing how acoustic siren recognition can preempt a signal cycle. The hard engineering problem is not lighting LEDs — it is recognizing a siren reliably. Ambulance sirens alternate between two tones (typically around 650 Hz and 950 Hz, a design reference used here), and the firmware looks for exactly that alternation in the FFT spectrum rather than reacting to loudness, which is what separates it from a toy that flips on every car horn. The override logic also matters: a real preemption must hold green long enough for the vehicle to clear the intersection and then release control safely with an amber transition and a hold-off timer after the siren fades.

1. The electret microphone module captures ambient sound; the ESP32's ADC samples it at approximately 8 kHz (a design target, 1024 samples per window).
2. Each window is passed through a 1024-point FFT on the ESP32, giving roughly 7.8 Hz frequency resolution — enough to separate the two siren tone bands.
3. The firmware measures energy in each tone band and checks whether the dominant tone alternates between them at the expected siren rate (each tone held approximately 0.4–0.8 s, a design value).
4. The alternation must persist across a configurable number of consecutive windows (design default 4) before detection is confirmed, rejecting single-tone horns and broadband noise.
5. On confirmation, the preemption state machine takes over the model intersection: cross lane to red, emergency lane through amber to green.
6. The state machine holds the preemption until the siren has been absent for the full hold-off time (design default 10 s), then releases control and resumes the normal cycle.
7. Every detection, release and timeout is logged over serial, so the demonstration shows exactly what the algorithm heard and decided.

Tech stack:

• ESP32-WROOM-32 development board
• Arduino IDE (ESP32 core)
• arduinoFFT library
• Electret microphone module with onboard amplifier (e.g. MAX4466-based)
• Model traffic light (3-LED, logic-level drive)
• Breadboard, jumper wires, 5 V USB supply
• Phone-played siren test audio (for calibration)

• [Two-tone siren recognition] The ESP32 samples microphone audio at approximately 8 kHz and runs a 1024-point FFT on-device, watching for energy that alternates between the two siren tone bands (design reference 650 Hz / 950 Hz) across a rolling detection window.
• [Signal preemption override] On confirmed siren detection, the firmware suspends the normal cycle, forces green for the emergency lane and red for the cross lane, then returns control through an amber transition.
• [Hold-off timer] A configurable hold-off (default design value 10 seconds) keeps the preemption active after the siren fades, so a momentarily quiet stretch of road noise does not release the green early.
• [Loudness-independent detection] The decision uses the alternating-tone pattern, not raw volume, so traffic noise and distance do not gate detection — only the siren signature does.
• [Tuneable detection thresholds] Siren tone frequencies, alternation timing, confirmation count and hold-off time are constants in the firmware, documented for viva discussion and adjustable during the buyer's own calibration.
• [Serial status log] Detection events, dominant tone, and preemption state changes print over USB serial for debugging and demonstration.
• [Model intersection controller] The ESP32 also drives the model intersection's normal red-amber-green cycle (all-LED, no mains wiring), which the preemption logic can interrupt and restore.
• [Calibration procedure] A step-by-step procedure to verify tone bands against a phone-played siren recording, set the confirmation count, and test false-trigger resistance with horn and music clips.

• Working prototype: ESP32 board, electret microphone module, model traffic light, wired on breadboard
• Complete firmware source code with commented detection and preemption state machine
• Circuit and wiring documentation with connection tables
• Component list with exact part descriptions
• Calibration procedure document (tone verification, threshold tuning, false-trigger testing)
• Project report PDF (background, siren acoustics, FFT design, methodology, results)
• PPT presentation for final review
• Viva Q&A preparation document (FFT, sampling, siren patterns, preemption logic, limitations)