Inspiration

Unmaintained industrial restrooms generate dangerous levels of ammonia NH3 gas from urine breakdown. Occupational safety standards set clear limits:25 PPM: NIOSH REL / ACGIH TLV (8-hour TWA limit).50 PPM: OSHA PEL (Permissible Exposure Limit for acute respiratory risk).Section 19 of the Factories Act, 1948 mandates factory sanitation, but plant managers rely on unverified paper logbooks. Existing cloud IoT monitors cost over ₹8,000 and fail in RF-shielded factory basements. We set out to build a sub-₹3,000, offline-first edge hub that replaces paper logs with chemical sensing, direct cellular alerts, and mathematical cleaning verification.

What it does

Monitors Air & Door Usage Offline: Tracks ammonia NH3 and stall access across 6 stalls without local Wi-Fi or cloud servers. Verifies Cleaning Math: Calculates chemical decay slope d(ppm)/dt) in C++ to confirm actual mopping while filtering ambient factory fumes. Automates Cellular Escalations: SIM7600E module places direct Stage 1 voice calls/SMS to cleaners at 25 PPM, escalating to Stage 2 supervisor calls at 50 PPM or after 15 minutes. Warns Workers Pre-Entry: Sounds a 95dB local alarm outside the block during severe gas spikes (>50 PPM) to deter entry into hazardous stalls. Logs Shifts Contactlessly: Janitors tap RFID cards to log shifts locally into onboard flash memory.

How we built it

Hardware: ESP32 edge microcontroller, CD4051 8-channel analog multiplexer (sharing 3× MQ-137 ammonia sensors across 6 stalls at a 1:2 ratio), 1× MQ-135 reference sensor, SIM7600E 4G LTE module, MFRC522 RFID reader, 6× LDR door sensors, and an active 95dB buzzer inside an IP65 enclosure. Storage: ESP32 SPIFFS flash memory with a 7-day circular buffer and NVS wear-leveling for critical escalation logs. Core Software Algorithm: The ESP32 calculates local chemical rate-of-change: Decay Velocity= d(ppm)/dt = d(PPM)/dt - PPM(t - Delta t)}{Delta t}A valid cleaning event is confirmed only when d(ppm)/dt < 0 after an RFID tap. A 24-hour dynamic baseline filters out background industrial solvents.

Challenges we ran into

Sensor Response Kinetic (T90): Fast 500ms multiplexer channel switching caused ghost readings due to the MQ-137's 30-second T90 recovery time. Solved by implementing firmware settling delays and signal averaging during calibration. False Alarms from Ambient Fumes: Solved by replacing static threshold triggers with derivative rate-of-change math d(ppm)/dt and dynamic rolling baselines. GSM Power Spikes: Added high-capacitance decoupling and dedicated power regulators to handle SIM7600E 2A current spikes during call placement without resetting the MCU.

Accomplishments that we're proud of

1.Low Cost: Achieved a ~₹2,150 COGS, enabling a ₹2,999 retail target—60%+ lower than cloud competitors. 2.100% Offline Resilience: Operates and alerts via direct cellular calls with zero reliance on cloud servers or internet. 3.Defensible Financials: Projected a realistic 3-year SOM of ₹15.0 Crore (25,000 stalls at ₹3,998 unit/Y1 AMC = ₹9.995 Cr + ₹5.00 Cr Y2–Y3 AMC).

What we learned

1.Accurate Claim Framing: Shifted from absolute claims like "zero data loss" to defensible phrasing: reducing network dependency and providing local data persistence. 2.Gas Sensor Dynamics: Learned to account for physical diffusion and sensor settling kinetics in code.

What's next for IoT-Based Washroom Hygiene, Compliance & Sterilization

1.Field Pilot (TRL 5/6): Build custom 2-layer PCBs and deploy pilot hubs across 3 active MSME plants within 6 months. 2.Patent Filing: File a provisional patent covering the derivative decay verification algorithm (d(ppm)/dt) and direct pre-entry audio warning workflow post-TRL 5 validation. 3.Automated Disinfection: Integrate relay outputs to trigger local UV-C or aerosol misting during non-occupancy hours.