The Living Box

The Living Box’s objective is to create the perfect environment for growing living organisms reducing the unpredictability of Nature.

The main challenges currently faced by the World and the agriculture industry are [1]:

1. Increase of 70% of demand for food;
2. 25% of farmland is already rated as degraded;
3. Water is highly stressed, 40% of the world´s rural population lives in water-scarce areas;
4. 80% of the global deforestation is driven by agricultural practices;
5. Greenhouse emissions doubled in the past 50 years.

Automated data-driven farming aims at optimal growing conditions, where harvests quality is maximized and resources consumption and costs are minimized.

The Living Box is an automated and data-driven green house.

Our vision is to implement sustainable and cost-efficient green houses to grow fresh produce while preserving both quality and the environment. Our green house is:

• Data-driven: Data power optimal and reproducible environments where crops can flourish;
• Cost efficient: Automation minimizes risk of losing crops as well as resources consumption, space, and labor.

Our green house brings value to farmers, farm-to-table restaurants, organic grocery stores, amateur growers, florists, mushroom growers, cannabis industry, beekeepers, among others.

The Living Box

Concept

The Living Box is composed by three groups of elements: sensors, a brain, and actuators. All together, sensors, the brain, and actuators cooperate and create the optimal environment for growing living organism.

The sensors measure environmental changes, such as temperature or humidity, and provide data to the brain.
The brain interprets the data read by the sensors, takes decisions, and triggers the actuators.
The actuators respond to the brain’s commands and modify the environment.

Prototype

The Living Box prototype and its components.

The Living Box’s current brain comprises a scheduler for period tasks (such as data collection or air ventilation) and PID controllers (for air humidity control). We are deploying neural networks for plant detection, plant size estimation, and diseases detection [2]. The brain runs on a Raspberry Pi which allows the Living Box to be cheap, independent, and remotely accessible.

The current sensors are: an air thermometer, a humidity sensor, a soil moisture sensor, and a camera. The data is periodically sent to the brain. For instance, the air temperature is measured every 5 minutes, while once a day the camera takes a picture of the growing crop.

The current actuators are: an air humidifier, two fans for air, a moisture circulation, and a water pump for irrigation.

Automated Data-Driven Growing

Following is an example of the Living Box autonomously cultivating basil. After setting humidity and air circulation thresholds, the Living Box autonomously controlled the environment in which the basil grew with no human intervention for over two months.

Basil germination. The neural network detects the basil’s sprouts.

The Living Box constantly collects data that is remotely accessible via a dashboard. The dashboard provides information from sensors and actuators such as images, temperature, humidity, or state of fan and humidifier.

Temperature and humidity (blue) trends and fan and humidifier states (red) trends. In this example, the Living Box autonomously triggers the humidifier to keep the humidity above 90% and periodically turns on the fan for ventilation.

Autonomously cultivated basil with no human intervention for over two months.

References

[1] 2018, The report “Agriculture 4.0”, launched in The World Government Summit.
[2] Demo: Grapevine Peronospora deterction https://www.youtube.com/watch?v=m3K-ouicZ_Y&t=4s