AGRILITI
October 23, 2022, Kraków
We started Agriliti to help our friends from Łąki Kwietne to measure and better understand the impact of sustainable agriculture on the environment, especially water conditions in cities. Initially the goal was to compare parameters in usual lawns to these with specially selected flowers and perennials. Later it evolved to a more generic IoT platform that allows it to gather data from various sensors and even video streaming. Our business use cases were also extended to general agriculture data gathering and processing.
Cloud architecture
Key assumption we’ve made was that we need to be able to gather data from multiple types of sensors, both supporting advanced communication stacks such as HTTPS and those with basic TCP connections. Another assumption was to create as simple as possible server side application to be able to fully focus on the embedded part as it was identified as the most complex. This led to architecture built based on InfluxDB and Grafana with Java Spring as the middle layer. Later we added a React web interface to allow device management and viewing real-time data streams from the devices.
Whole application is packed inside Docker images and running behind a Nginx proxy. Nginx is also hosting React app builds. Thanks to this approach all traffic is easily encrypted with SSL for both HTTP and TCP requests.
After a few years of data collection we can admit that this architecture allows quite flexible data modeling and is very robust. Difficulties with real-time data presentation can be addressed with dedicated Spring and React features that come with a little bit of cost, but usually it is worth it!
Devices
During the project we’ve come up with two major types of sensors that are reporting the data. The main concerns were how devices are powered and how data is transmitted to the cloud:
-
Pure embedded device with solar panel as energy source with its own Radio Access modem
-
Raspberry Pi device powered from the grid with Radio Access modem
-
RB Pi was used as LoRa relay for tiny devices that are battery powered
-
Solar powered devices must be super energy efficient so it is using a 2G modem and transmits the data every single hour. Core of it is an STM microcontroller that communicates with sensors and modem. Device is also equipped with a battery and charging section. It can last more than 2 days without sunlight. If it happens to be very dark for a longer period of time it can automatically wakes up when sun comes back and start collecting the data again.
More advanced deployment consists of grid powered Raspberry Pi Zero that can handle multiple sensors and LoRa transceiver. Multiple soil parameters sensors are dug underground, battery powered and again controlled by STM microcontroller. This creates a star topology in the measurement field. Thanks to unlimited power supply for RB, LTE modem can be used to deliver real time video stream or movement detection. On the other hand small LoRa devices are energy efficient and can stand more then 6 months without charging.
Deployments so far
Wody Polskie - Bydgoszcz
In this deployment our goal was to prove that soil temperature and humidity have better levels in areas below wild flowers meadow. Two solar powered devices are used in the Wody Polskie facility in Bydgoszcz. Slight disadvantage of this solution is that the mounted device makes it hard to mow the grass and also when flowers meadow is too high, it might disturb energy harvesting by solar panel. Disregarding these two problems, devices were measuring soil and air parameters for more than a year and the data was used by regulators to justify flower meadow positive impact on hydro-ecosystem.
Instytut Ogrodnictwa - Skierniewice
In this deployment we’ve combined Raspberry and battery powered tiny sensors communicating via LoRa. This setup is used in Instytut Ogrodnictwa in Skierniewice and its goal is to provide data about impacts of low crops on fruit trees. Battery powered devices are deployed all over the facility and are reporting data every 10 minutes to the Main Station which is forwarding measurements to the cloud. With this approach our equipment does not impact caring of crops, but these tiny devices have to be charged once per season. Additionally with quite advanced Raspberry based Main Station we can provide on-demand, live video stream and movement detection functionality.
Summary
During the project we’ve successfully combined embedded sensors, cloud infrastructure and web application. This led us to the creation of complete Internet of Things system deployment where it is not only possible to gather measurements, but also to interact with devices in real time. The most important thing for us is that our work is used to find the best ways to protect the environment. We are proud we can use our technology skills on the journey for the solutions to the most challenging problem in human history.
Team
