IoT as an enabler of AI
Solar energy, water, algae, and plankton – that’s all that Blue Planet Ecosystems needs to breed fish in a completely automated process. Cloudflight offers a solution from hardware to application, thereby enabling end-to-end steps in all the phases – from design, build and operate phases. With IoT (a hardware/software ecosystem), it is remotely monitored and controlled, and continuously optimized in the process.
- Blue Planet Ecosystems is a start–up and needs to act quickly
- The customer has changing requirements
- There is no vendor lock-in; independence
Main question: how to obtain IoT data as platform-independent as possible and extensible for anyone who needs to use it.
Cloudflight has led two projects for Blue Planet Ecosystems. The first challenge was that the client already had control systems but was using Raspberry Pi and WIFI; the system often failed, and data was lost.
The goal was to have a fail-safe control system that can be relied upon.
The second project was about how to handle data. The Cloudflight team prepared an IoT platform approach. We didn’t want to write software that simply receives and transmits data – because there are already relatively many ready-made platforms.
The idea of BPE is to create a source of sustainable seafood in urban and desert environments through a data-driven approach. Cloudflight is helping BPE implement the necessary cloud-based/edge infrastructure and software needed to achieve this.
In Project 1, Cloudflight’s idea was to build its own IoT Platform.
What is IoT?
IoT describes a hardware/software ecosystem that includes:
- the “things” side, which typically consists of network-enabled sensors (e.g., light, temperature, and distance) and actuators (e.g., motors, servos, heaters, and pumps)
- a platform that allows the management of these “things” and the data they generate by forwarding user applications that generate value from the data thus collected (e.g., dashboards for visualization or machine learning models for predictive maintenance). Insights generated in this way can be translated back into input for connected things. For example, new set–point temperatures that switch over a heating system
The idea of BPE is to:
- use minimal amounts of water (run recycling, detect leaks)
- use solar power when available (solar + grid power battery; the hardware runs off battery)
- optimize energy-intensive operations
We want to collect and process data that accurately represents all the different processes going on in the tanks so that we can build up enough data to train an AI that can then optimize those processes.
Another idea is to create two control circuits. The first control circuit observes the temperature in the tank. The control boxes can then turn on the heating – without a computer.
Using video cameras, the fish can be recorded to make sure that they are doing well. This machine learning and AI aspect should happen in an extended circle – so that it can be noticed in time whether the fish are sick.
Another idea is that device management will be implemented using cloud services.
Future outlook: to create a web-based app based in the cloud. This app should present historical and predicted sensor data to biologists, who will then be able to act and send back commands to the control units.
Cloudflight’s solution is to enable the entire process from start to finish. We build the premise and the foundation for it in order to ensure that the data is both complete and accurate. Cloudflight has built a system that is versatile; it can connect either via WIFI or via mobile networks.
Our motto is: think along and prepare the whole process from the sensor to the web application with data.
Are you curious about how to revolutionize fish production? You can learn more about it in the Data Scientist Blog written by Lukas Winkler.
Blue Planet Ecosystems is a biotech start-up that wants to revolutionize food production methods. In their artificial ecosystems, they produce fish from sunlight. In this concept, algae, water, sunlight, and IoT support fish farming.
In the current project, Cloudflight is supporting the second version of BPE’s control system by applying our IoT expertise and thus improving their processes and the general reliability of the system. We are also connecting this new hardware with the AWS IoT Core and preparing the creation of a new web-based app for controlling the farms.