General details
EDIHs involved
Challenges
Paintec lends its technical expertise to the Food and Agriculture Organization (FAO) – a specialized agency of the United Nations – in a project dedicated to identifying vegetation zones with a high likelihood of locust swarm development based on their unique characteristics. The ultimate objective is to fumigate the most potentially hazardous areas for crops within the controlled zones.
For detection purposes, entirely autonomous drones are employed, requiring minimal or no specialized training for operators. These drones are designed to operate in harsh environments such as deserts, demanding minimal maintenance that can be handled by the personnel responsible for each unit. While cost-effectiveness is a key consideration, the system must also enable the export of a portion of the captured information to the cloud from the drone's base station or from network connection points.
The challenge was to upgrade the hardware (drone control board, on-board computer, batteries, component architecture, etc.) and software (drone system architecture, cloud platform, ML/DL algorithms for vegetation detection, etc.). Component supply chain problems were an additional challenge. An architecture was required to enable easy and compatible construction and assembly of different components.
Solutions
A series of tasks were undertaken to address the challenges faced by Paintec and to build upon the existing platform, which comprises a drone with a flight controller, edge computer, and user interface, and a cloud platform. The initial platform was a proof of concept developed by external developers working for Paintec.
The first iteration followed Paintec’s meticulous definition of the objectives and priorities for each identified module. This comprehensive approach led to a thorough analysis of both hardware and software components.
The software components, the drone's control unit (implemented on a Raspberry Pi 4), and the DOMA platform (a repository for uploading and reporting information extracted from captured images), were reengineered, refactored, containerized (dockerized), and documented. Computing performance tests were carried out to determine the restrictions to the drone’s control unit that should be considered by future improvements (i.e. software library versions).
The hardware components were carefully examined to identify potential upgrades or alternatives. IoT device analysis focused on the battery, flight control unit, and computing hardware, and particularly Raspberry Pi devices. The goal was to identify alternative components that would enable the integration of new drones while adhering to FAO constraints. Additionally, a hardware architecture was devised to streamline the assembly process undertaken by Paintec.
After a few iterations, a clear hardware/software architecture was obtained which allows Paintec to be more resilient to hardware pressures. In addition, the architecture and the performance test results enabled Painted to establish a roadmap for the improvement of drone platforms, in compliance with FAO and customer requirements.
Results and Benefits
The main results achieved by the service/project were:
-
A hardware and software architecture which allows Paintec to identify the implemented components, their internal and external interfaces, and interactions:
-
A documented methodology for the deployment of the full platform in new customers.
-
A platform improvement roadmap;
-
Specifications of hardware connections;
-
-
A set of alternative hardware components which can be used to replace those currently used with minor or no software customizations;
-
A benchmark of the computing capabilities required to support new ML models which improve current capabilities to detect vegetation;
-
New ML models to detect vegetation more accurately.
Following the end of the project, Paintec is more resilient to potential shortages of hardware components, and are more capable of industrializing the construction of the drones. In addition, their software quality has increased, as recurrent deployment can now be performed in an automated way, both with in the drone controlling subsystem and on the cloud user interface subsystem. Moreover, the architecture, the benchmark and the roadmap of new improvements assures the quality of the necessary upgrades for future platform enhancements.
Perceived social/economic impact
Pest monitoring and control is a crucial aspect of sustainable agriculture, particularly in developing countries where farmers rely heavily on their crops for sustenance. Achieving Sustainable Development Goals (SDGs) hinges on effective pest management strategies that not only safeguard food security but also minimize environmental harm and promote sustainable farming practices. Although pest management poses a range of challenges in developed nations, it is an absolute necessity in developing regions, where Paintec's drones are making a significant impact.
Through the solution provided for Paintec, the social impact of the drones became very clear. These drones empower farmers and authorities in developing countries to effectively monitor the riskier areas for the development of locust swarm.
Paintec's efforts could mark the beginning of a promising future where locally manufactured drones, designed specifically for the needs of developing regions, revolutionize pest monitoring and control. By tailoring these drones to the unique environmental and technological constraints of these areas, Paintec has the potential to expand the reach of its technology and amplify its positive impact on global food security and sustainable agriculture.
Measurable data
Not yet available.
DMA score and results - Stage 0
Paintec’s T0 DMA Global Score was 33%.
Their scores per DMA segment were:
-
Digital Business Strategy – 37%
-
Digital Readiness – 62%
-
Human-Centric Digitalisation – 13%
-
Data Governance – 23%
-
Automation & Artificial Intelligence – 40%
-
Green Digitalisation – 20%
There are huge possibilities for improvement, that will be faced in different stages, as the company is new and they are growing really fast.
DMA score and results – Stage 1
Not yet available.
Lessons learned
Handling the expectations of a company is not easy when we are talking about a tech SME. The time to market of any company is very short, but for this kind of SME, it is even more pressurized. At ITA (Aragon EDIH service provider), we adapted the CRISP-DM methodology, which is considered the de facto ML project methodology, to guide our projects. This iterative methodology helps to build better knowledge and understanding about company challenges, the data required and its business value.
In our experience, it is better to start by solving the easiest problems and/or implementing a basic solution, before moving on to a more complex problems/solutions. An iterative process that involves regular follow up meetings enables the detection of deviations, the communication of errors and agreement on solutions and/or the selected alternatives. In addition, short iterations (with a duration of 1 or 2 weeks) make it possible to re-prioritize the next actions and adapt them to the SME’s changing reality.
Need support?
Consult our catalogue to locate the Eupopean Digital Innovation Hub nearest to you and accelerate your company's digital transformation.