We're thrilled to announce that SafeHat WorkNet won the IoT Hardware Hack award at TAMUHACK25!
| Member | Role & Contributions | |
|---|---|---|
| Cesar Magana | Software Engineer - ESP32 Mesh Network & Data Serialization | LinkedIn Profile |
| Brighton Sikarskie | Hardware Engineer - Sensor Integration & Mesh Network Development | LinkedIn Profile |
| Samuel Bush | Software Engineer - Server & Dashboard Development | LinkedIn Profile |
Project Overview
SafeHat WorkNet is an innovative safety helmet system designed to enhance worker protection in hazardous environments like construction sites, industrial plants, and mines. This project leverages the power of IoT, a self-healing ESP32-based mesh network, and real-time data monitoring to create a comprehensive safety solution that is easily deployable using a portable Raspberry Pi network.
Our Mission
- [Specific Problem]: Address a critical safety challenge faced in the target industry (e.g., lack of real-time hazard awareness, delayed emergency response, etc.).
- [Solution Approach]: SafeHat WorkNet provides real-time hazard alerts and tracking using a self-healing mesh network for rapid response capabilities. Emphasize the hackathon's theme if applicable.
- [Impact]: Highlight the potential positive impact on worker safety, incident prevention, and industry standards. We aim to reduce workplace accidents and improve emergency response times.
- Temperature and Humidity (DHT22): Monitor worksite conditions to prevent heat stress.
- Gas Sensor (MQ135): Detect hazardous gas levels (CO2, etc.) with configurable thresholds.
- Light Sensor (BH1750): Assess ambient light for visibility optimization.
- Accelerometer and Gyroscope (MPU6050): Detect falls, impacts, or irregular movements, triggering alerts.
- Magnetometer (QMC5883L): Provides heading information.
- Robust, Decentralized Communication: Creates a reliable network between helmets, ensuring continuous data flow even if some nodes fail. (H2H - Helmet-to-Helmet Communication!)
- Fault Tolerance: Automatically reroutes data if a node goes offline, maintaining network integrity.
- Dynamic Root Node Selection: Intelligently elects a "root node" based on signal strength (RSSI). The root node is likely responsible for relaying data within the mesh. If the root node fails or its signal weakens, a new root node is dynamically elected.
- Scalability: Easily add more helmets to the network.
- Bridge Node Election: Dynamically selects a node with optimal server connectivity to act as a gateway to the Raspberry Pi server.
- Custom Communication Protocol: We developed a custom JSON serialization protocol for efficient data transfer between nodes and the server. This protocol includes RSSI values and sensor readings, enabling location approximation and optimized for low-bandwidth mesh networks. Each node broadcasts its data, which is collected and processed by the dynamically elected root node.
- Onboard LED: Visual indicator of alerts and status.
- Centralized Dashboard: (See "Dashboard" section below) Provides a live overview of worker status, environmental conditions, and approximate locations.
- Server Communication: Bridge nodes relay data to a central server (Flask-based) for data logging, analysis, and rapid response coordination.
- Database Storage (SQLite): Stores sensor data, events, and location approximations for analysis.
- API Endpoints: Enables data retrieval for monitoring and integration with other systems.
- Real-time Visualization: Displays sensor data in interactive charts (Chart.js) for easy monitoring.
- Node Status Overview: Tracks the status of each helmet/node in the mesh network.
- Event Log: Records alerts, warnings, and other significant events.
- Portable Server: The system is designed to work with a portable Raspberry Pi server, making it ideal for temporary worksites or locations with limited infrastructure.
- Local Operation: Operates independently on a local network, enhancing data security and eliminating the need for constant external network access.
- Construction Sites: Enhance safety by monitoring environmental hazards, worker movements, and enabling rapid emergency response.
- Industrial Plants: Provide real-time alerts and monitoring in high-risk areas.
- Mining Operations: Improve safety in confined spaces with gas detection, fall monitoring, and location approximation.
- Emergency Response: Equip first responders with SafeHat for situational awareness and improved coordination.
- Incident Analysis: Review historical data to understand incidents and improve safety protocols.
- Hardware: ESP32 Microcontrollers, DHT22, MQ135, BH1750, MPU6050, QMC5883L sensors, Raspberry Pi (for server).
- Firmware: C++ (PlatformIO), painlessMesh library.
- Server: Python (Flask), SQLAlchemy (for database interaction).
- Dashboard: HTML, CSS, JavaScript, Chart.js.
- Database: SQLite.
- Network: Ad-hoc mesh network (ESP32), Wi-Fi (for Raspberry Pi server).
- Sensor Drivers: Custom-built using
esp-idf(not Arduino libraries).
- [Challenge 1]: Achieving reliable communication in a dynamic, potentially obstructed environment.
- [Solution]: Implemented a self-healing mesh network using ESP32s and the painlessMesh library, featuring dynamic root node selection and H2H communication.
- [Challenge 2]: Real-time data processing, visualization, and location approximation.
- [Solution]: Developed a Flask-based server with API endpoints, an interactive dashboard using Chart.js, and implemented location approximation using the Shannon-Hartley theorem based on RSSI values.
- [Challenge 3]: Power optimization for extended use.
- [Solution]: (Describe the power-saving strategies you implemented or planned to implement, e.g., optimized data transmission intervals, sleep modes).
- [Challenge 4]: Ensuring accurate sensor readings in a noisy environment.
- [Solution]: (Describe any calibration or filtering techniques you used, e.g., calibration of the MQ135 sensor, filtering of accelerometer data).
- [Challenge]: Developing efficient and reliable communication on a resource-constrained mesh network.
- [Solution]: Created a custom JSON serialization protocol for data transfer, optimizing bandwidth usage and enabling features like RSSI-based location approximation.
- [Challenge]: Implementing sensor drivers from the ground up for greater control and optimization.
- [Solution]: Developed custom sensor drivers using
esp-idf, bypassing the need for potentially less efficient Arduino libraries.
- [Solution]: Developed custom sensor drivers using
SafeHat WorkNet is designed to provide approximate location information for each helmet, even without GPS. This is achieved through the following:
- RSSI-Based Triangulation: The system leverages the Received Signal Strength Indicator (RSSI) values between helmets in the mesh network. Stronger RSSI values generally indicate closer proximity.
- Shannon-Hartley Theorem: We plan to incorporate the Shannon-Hartley theorem to further refine location estimates by considering factors like signal-to-noise ratio.
- Localized Alert Propagation: A key application of location approximation is to enable localized alerts. When a hazard is detected by a helmet (e.g., high gas concentration), the alert can be propagated to other helmets within a certain RSSI range, providing immediate warnings to nearby workers.
- GPS Integration: Add GPS modules for precise location tracking.
- Onboard Communication: Integrate speakers and microphones for two-way voice communication between helmets.
- Further Sensor Integration: Expand the range of sensors to monitor other environmental factors or worker biometrics.
- Machine Learning: Implement ML models for predictive hazard analysis, personalized safety recommendations, and improved location accuracy.
- Cloud Connectivity: Migrate the server to a cloud platform for scalability, remote access, and data backup.
- Dashboard Visualization: Visualize the approximate locations of helmets on the dashboard, providing a spatial overview of the workforce.
This project is licensed under the MIT License. See the LICENSE file for more details.
- ESP32 Community and painlessMesh library developers.
- Open-source sensor library contributors.
