Plan Bee Labs

Problems

Existing business models within the current landscape of remote honeybee hive monitoring predominantly revolve around a singular stakeholder, namely the beekeeper. These models are largely centralized and typically hinge on either hardware sales and installation or subscription-based services. However, beekeepers commonly exhibit hesitancy to invest in such systems and services, leading to limited market penetration. Consequently, even if the reliability of the collected data is guaranteed (and in many cases is not), its impact on conservation efforts remains constrained.

Proposed Solution

In our opinion, the main stakeholders should be crop growers and agroecosystems. Utilizing the implementation of a fully decentralized network and marketplace emerges as a promising solution to address this challenge. Enabling individuals to operate a node equipped with nesting site infrastructure and incentivizing their participation by contributing or selling collected trustless datasets on a decentralized marketplace creates a more inclusive model. Moreover, the opportunity for users to develop applications that leverage existing datasets on the marketplace to produce more valuable insights further enhances acceptance and engagement among interested parties.

Current state

The current state of the project is as follows:

• Bumblebees; Complete software system and hardware prototypes (version 1, pre-MVP) have been developed, and initial field tests have been successfully conducted. Version 2 is currently under development, incorporating insights gained from the field tests.
• Honeybees; Similar to the bumblebee project, complete software system and hardware prototypes (version 1, pre-MVP) have been developed, with successful initial field tests. Version 2 is also in development, building upon the lessons learned from the field tests.
• Wild pollinators. This project is in its early stages, focusing on planning, architecture construction, prototype development, and formulation of predictive models. Progress so far includes:
  • Training of the first Yolov8 custom model using open iNaturalist datasets, enabling prediction and recognition of various species such as Apis cerana, Apis dorsata, Apis florea, Apis mellifera, and several species of Bombus.
  • Programming and successful testing of the ESP32 module integrated with a camera against the trained model.
  • Development of foundational elements for the full-scale node on libP2P, facilitating peering between nodes and operating a YOLOv8 prediction model, including the exchange of service catalogues containing dataset and service information.
  • In progress: development of a mini node prototype utilizing the NVIDIA Jetson Nano Developer Kit for on-site operations in locations with reliable power supply.
  • Initiation of the development process for a desktop application aimed at facilitating the operation of a full-scale node on a personal computer.

Milestones

• Beginning July 2024, Full scale node v1.0.0
• Mid July 2024, Yolov8 custom model trained with most relevant bee pollinators (v1.0.0)
• Beginning August 2024, Firmwares v1.0.0 (dev boards)
• Mid August 2024, Tests with development boards
• Beginning September 2024, 1st nesting site / bee hotel prototypes ready
• Beginning October 2024, Yolov8 custom model trained (v2.0.0)
• Mid November 2024, PCB designs ready v1.0.0
• End November 2024, Firmwares v1.0.0
• End December 2024, 1st PCB prototypes ready
• End January 2025, In-house testing
• End February 2025, Release 1
• End April 2025, Outdoor testing (prototypes)
• May 2025, Production ready

Reference Documents

• Website (1)
• Website (2)
• Whitepaper
• Presentation