SmartBees

In Slovenia and many other parts of the world, numerous wild bee species thrive, each with its unique habits and adaptations. These tireless pollinators play a crucial role in our ecosystems, ensuring the growth of crops and maintaining biodiversity.

Our innovative project aims to monitor these vital insects by creating a smart bee home equipped with microphones and sensors. Through the power of machine learning, we analyze intricate bee behavior patterns to gain valuable insights into their environment.

Join us on this buzzing journey as we construct a bridge between technology and nature. 🐝🌼

Hardware

The first step of the hardware development involved a careful selection of components for our PCB. Initially, we considered a microcontroller board with integrated sensors. However, we soon realized that designing our own PCB with the ESP32-PICO-MINI-02 microprocessor would be more effective. This allowed us to add specific sensors required for comprehensive bee monitoring.

Additionally, we ensured that all sensors had Arduino support libraries, simplifying the firmware development process.

We designed a power management system that is both efficient and user-friendly. The system is powered by Energizer 1.5V AAA Lithium Batteries, chosen for their reliability, long shelf life, and ease of replacement. These batteries are housed in a Keystone battery holder, which holds three AAA batteries in series, providing a total of 4.5V. This is directly connected to our PCB.

For the hardware schematic and layout, we used the EasyEDA editor. The schematic was a grouped selection of the parts we decided to use and the passive elements that go along with them. The layout had a couple of interesting features, such as the cutout where the Bosch sensor and a microphone are positioned so that they are somewhat isolated from the rest of the circuit, and the 2.54mm pin headers that connect the main PCB to the secondary one that is used as an additional microphone.

Our two PCBs needed protection from outside factors, so we decided to put them in plastic housing. We searched for the housing that had the closest dimensions to our PCBs. As our PCB was being modified throughout the building process and changing size, we also needed to change the housing several times. When the PCB had its last version, we took the datasheets from the housings and resized our PCBs to fit in and made holes so we could secure the PCB. With the help of Inventor, a 3D building program, we checked if our PCBs and the housings were compatible.

After gathering all necessary elements and designing the circuit board, we needed to create a Bill of Materials (BOM). This document included every component, its availability, and the price per unit. We also incorporated all essential information required for a smooth project flow, including specifications and technical details. Creating this BOM was crucial as it provided a clear and organized overview of the required parts. This ensured we could efficiently and economically procure all components while keeping costs within budget.

Firmware

Firmware team's objective was to bridge inputs and outputs. Using ESP32 we read from sensors, and output to Lora via AT commands. To acomplish this we used RTOS and I2C protocol for data transmission, and I2S for reading from the microphone. Packages to Lora are sent in "packed" structure. Additionally we implemented a template debugging function for sent values in our software. In the next step, we will be implementing a sleep function for low power operation.

Networking

While working on the SmartBees project, the Networking team focused on getting the data from the Smart Bee home to the end user. To accomplish that goal we used the LoRa gateway platform TheThingsNetwork, which was used to get the sensor data from a remote location to the Internet. For parsing and displaying the sensor data we utilized Thingsboard, which proved to be an excellent choice due to its flexibility and intuitive design. This networking chain allows the sensor data to be viewed on an easy to use web based dashboard.

Machine learning with Edge impulse

As part of the Smart Bee Homes project, the machine learning team focused on developing a model to detect the presence of bees using audio data. We received 4 second sound clips, labeled as "buzz" or "nobuzz". We utilized Edge Impulse to test various machine learning models. Our goal was to create a model that accurately identifies bee activity through sound, enabling real-time monitoring and insights into hive health and behavior.

User interface

To enhance user experience, we aim to develop a user-friendly web application featuring an interactive dashboard. This application will efficiently manage and showcase real-time data in an organized manner, ensuring users can easily access and interpret the information they need. By offering valuable insights and project updates directly through the app interface, we ensure that users remain informed and engaged, facilitating better decision-making and project management.