IoT Restroom Occupancy Indicator with Usage Analytics

In malls, colleges, offices and transit hubs, queues form outside washrooms because visitors cannot tell which stalls are free — they wait at the entrance, push half-open doors, or walk away. Facility teams face the mirrored problem: cleaning rounds run on a fixed schedule, so high-traffic stalls get dirty between rounds while low-use ones are cleaned needlessly, and nobody has usage data to plan better. This project fixes both with per-stall occupancy sensing. A door-latch switch plus an IR proximity sensor reports each stall's state; an ESP32 gateway drives red/green indicators above every stall door and a corridor display showing the live free-stall count; and a web dashboard turns the stream into usage analytics — per-stall door-cycle counts, hourly occupancy patterns, peak-hour detection — with cleaning-cycle alerts when a stall crosses a configurable usage threshold. Critically, the sensing is occupancy-only: no cameras, no audio, no identity data. Privacy holds by architecture rather than policy, so the system is genuinely deployable in real institutional washrooms rather than just demoable in a lab.

1. The reed switch on each stall latch reports locked/unlocked while the IR proximity sensor checks for a person inside.
2. The ESP32 stall node fuses both signals into an occupied/vacant state with debounce (design target ~3 s) so door bumps don't flip the state.
3. The node drives the red/green LED above that stall's door and publishes the state change over Wi-Fi (MQTT) to the backend.
4. The corridor display panel updates the live free-stall count within seconds of every state change.
5. The backend logs each door-cycle event with a timestamp into the usage store.
6. The dashboard renders the live occupancy grid, per-stall usage bars, hourly occupancy charts and the peak-hour window.
7. When a stall's usage count since its last clean crosses the configured threshold (default 40, configurable), a cleaning-cycle alert appears on the dashboard until the stall is marked cleaned.

Tech stack:

• ESP32 (stall nodes + gateway, firmware in Arduino/ESP-IDF style C++)
• Reed switch · IR proximity sensor
• Red/green LED stall indicator panels
• Corridor display (MAX7219 LED matrix / OLED)
• MQTT · Wi-Fi telemetry
• Node.js backend with event store
• HTML · CSS · JS dashboard
• 5 V DC power adapters

• [Per-stall door-latch sensing] A reed switch on each stall's latch reports locked/unlocked state — the primary, positive occupancy signal straight from the door.
• [IR presence backup] An IR proximity sensor backs up the latch state, so an unlocked-but-occupied stall is not misread as free; both signals fuse with debounce.
• [Stall and corridor indicators] Red/green LEDs above each stall door plus a corridor panel showing the live free-stall count, so visitors see availability before queuing.
• [Live occupancy dashboard] A web dashboard mirrors every stall's state in real time over Wi-Fi/MQTT, with per-stall state history for the shift.
• [Usage analytics] Door-cycle counts per stall, hourly occupancy charts and peak-hour detection turn raw events into facility-usable insight.
• [Cleaning-cycle alerts] When a stall's usage count since its last clean crosses the configurable threshold, the dashboard raises an alert for the cleaning team; marking the stall cleaned resets its counter.
• [Privacy-first design] Occupancy-only sensing — no cameras, no audio, no identity data — so the system is acceptable in real washrooms, not just labs.

• Stall sensor modules: reed latch switch + IR proximity sensor with wiring per stall
• ESP32 node/gateway with firmware (sensing fusion, debounce, MQTT, display driver)
• Assembled corridor display panel (LED matrix/OLED) with mounting bracket
• Web dashboard with live occupancy grid, usage analytics and cleaning alerts
• Installation and door-mounting guide with sensor calibration procedure
• Project report PDF (sensing design, system architecture, methodology)
• PPT presentation and viva Q&A preparation document

• Occupancy-only: the system reports stall states, not people — it cannot confirm the type of usage or count multiple occupants in one stall.
• No cameras by design, so analytics are limited to door-cycle counts and occupancy durations; no demographic or behavioural insight is possible.
• IR sensors can be affected by highly reflective surfaces or dust build-up; the installation guide includes a placement and verification procedure to run during calibration.
• Latch switches depend on correct door alignment — poorly aligned latches must be adjusted before sensing is reliable.
• Wi-Fi coverage is required across the washroom block for live updates; outages are bridged by store-and-forward (design target), not eliminated.
• Cleaning alerts are usage-based, not hygiene-based: a spill in a low-use stall still needs a manual report.
• Real public deployment should include signage that occupancy is monitored, and local privacy regulations should be checked.