NexusMesh 🏆

Best Hardware Hack — HackKU 2026

A distributed, AI-powered eco-physiological sensing platform that protects workers and communities in real time — no cloud, no delay, no privacy concerns.

What It Does

NexusMesh is a three-node wireless sensor network built on ESP32 microcontrollers. A central sender node reads body temperature, heart rate, and gas presence simultaneously, then broadcasts all data over ESP-NOW (Wi-Fi peer-to-peer, no router required) to two receivers: • A handheld display node shows live readings on a 0.96" OLED — portable enough for a field worker to carry. • An AI analysis node connected to a computer runs a multi-user anomaly detection engine that learns each individual's personal temperature baseline and triggers escalating alerts when deviations, rapid changes, or gas events are detected. All inference runs locally on the ESP32. Zero cloud dependency. Architecture ┌─────────────────────────────────┐ │ SENDER NODE │ │ ESP32-C6 │ │ ├── MLX90614 (IR temperature) │ │ ├── MAX30102 (heart rate/SpO2)│ │ └── MQ-135 (hazardous gas) │ │ │ │ Broadcasts via ESP-NOW @ 1Hz │ └────────────┬────────────────────┘ │ ESP-NOW (802.11, no router) ┌───────┴────────┐ │ │ ▼ ▼ ┌──────────────┐ ┌────────────────────────────────┐ │ HANDHELD │ │ AI RECEIVER NODE │ │ NODE │ │ ESP32-C6 │ │ ESP32 │ │ │ │ 0.96" OLED │ │ MultiUserTempAI engine: │ │ │ │ • Personal baseline learning │ │ Displays: │ │ • 4-tier anomaly detection │ │ • Temp °F │ │ • Rapid change detection │ │ • BPM │ │ • Fever progression tracking │ │ • Gas alert │ │ • Gas exposure history │ │ • Finger │ │ • Up to 10 concurrent users │ │ detected │ │ • MAC-based user identity │ └──────────────┘ └────────────────────────────────┘ Hardware Component Purpose Interface ESP32-C6 (×3) Main MCU for all three nodes — MLX90614 Non-contact IR body temperature I2C (SDA: GPIO4, SCL: GPIO5) MAX30102 Heart rate & SpO2 I2C (SDA: GPIO6, SCL: GPIO7) MQ-135 Hazardous gas detection Digital (GPIO22) SSD1306 0.96" OLED Live data display (handheld node) I2C Note: The sender uses two separate I2C buses — Wire for the MLX90614 and Wire1 for the MAX30102 — to avoid address conflicts. Files File Node Description esp32_sender.ino Sender Reads all sensors, broadcasts struct via ESP-NOW to two receivers every ~1s esp32_handheld_node.ino Handheld Receives data, displays temp/BPM/gas status on OLED ai_receiver_node.ino AI Node Full anomaly detection engine, outputs to Serial Monitor AI Anomaly Detection The AI node (ai_receiver_node.ino) implements a MultiUserTempAI class that supports up to 10 concurrent users, identified by their sender's MAC address. How the baseline learning works Each user's last 50 temperature readings are stored in a circular buffer. Once 20 readings are collected, the system computes a personal normal range (mean ± std dev) using only clinically normal readings (36–37.5°C). The baseline updates every 10 readings thereafter. Alert tiers (in priority order) Priority Trigger Alert Level 1 Gas detected (MQ-135 LOW) 🚨 EVACUATE — immediate action 2 Temp ≥ 39.5°C 🚨 CRITICAL FEVER — emergency 3 Temp ≥ 38.5°C ⚠ HIGH FEVER — contact supervisor 4 Temp ≥ 38°C ⚠ FEVER — rest and monitor 5 Temp ≤ 35°C ❄ HYPOTHERMIA — contact supervisor 6 > 1°C from personal baseline ⚠ ELEVATED / LOWERED 7 > 2× std dev from baseline ⚠ UNUSUAL pattern 8 > 0.8°C rapid change (5-reading window) ⚠ RAPID CHANGE 9 Rising trend above 37.5°C ⚠ FEVER STARTING — All clear ✅ NORMAL Gas exposures are tracked per user — three or more exposures triggers an additional medical attention recommendation. Serial commands (AI node) list → Show all registered users and their baselines stats → Heap, CPU freq, uptime, last data timestamp clear → Clear terminal (ANSI) help → Show command list Setup & Flashing Requirements • Arduino IDE 2.x or PlatformIO • ESP32 board package installed • Libraries: o Adafruit MLX90614 o Adafruit SSD1306 + Adafruit GFX o SparkFun MAX3010x Steps

1. 2. 3. Flash the AI receiver node first, open Serial Monitor at 115200 baud, and note its MAC address from the startup output. Flash the handheld node, note its MAC address. Update esp32_sender.ino with both receiver MAC addresses:
2. uint8_t receiver1MAC[] = {0xXX, 0xXX, 0xXX, 0xXX, 0xXX, 0xXX}; // Handhelduint8_t receiver2MAC[] = {0xXX, 0xXX, 0xXX, 0xXX, 0xXX, 0xXX}; // AI node
3. Flash the sender node.
4. 7. Allow 20 seconds for the MQ-135 to warm up before readings are valid. Open Serial Monitor on the AI node — data will start streaming immediately. Dependencies (included as submodules) • Adafruit_BusIO • Adafruit_GFX_Library • Adafruit_MLX90614_Library • Adafruit_SSD1306 • Adafruit_Unified_Sensor • SparkFun_MAX3010x_Pulse_and_Proximity_Sensor_Library Use Cases • Industrial safety — factory floors, chemical handling, confined space entry • Healthcare screening — fever detection without physical contact • Field worker monitoring — portable handheld node for remote environments • Research — personal physiological baseline tracking over time Built At HackKU 2026 — University of Kansas 🏆 Winner: Best Hardware Hack License MIT — free to use, modify, and build on.