AI Anomaly Detection: Transforming Industry Operations
2024-12-09
The efficient management of crops is crucial for ensuring food security and environmental sustainability. Agriculture faces numerous challenges, including water scarcity, climate change, and the pressure to increase productivity. Automated irrigation and AI-assisted crop management offer promising solutions to address these issues. Automated irrigation plays a crucial role in optimizing water use, which is one of the most important resources in agriculture. Smart irrigation systems enable the application of the right amount of water at the right time, leading to water savings, increased efficiency, improved crop yields, cost reduction, and greater sustainability.
Tupl is participating in an Innovation Project focused on the use of artificial intelligence for automatic and smart management irrigation of green areas. The main objective of this project is to enhance the management and maintenance of green areas by implementing advanced sensor, communication, and artificial intelligence technologies. The implementation process is being carried out by a consortium of companies, including UMA, INNOVA IRV, Smart City Cluster, Cordis SUITE, and Tupl. Tupl is providing its platform, TuplOS to facilitate and improve the analysis of field data and the development of action plans for farmers.
To develop this AI Based Innovation Project for automatic management and intelligent irrigation of green areas, the consortium, as mentioned earlier, was created between UMA, Cordis SUITE, INNOVA IRV, Tupl, and, as project leader, Smart City Cluster. The project "AI System for Automatic Green Area Management" has been funded under the Digital Technologies Projects line of the 2023 Call for Aid to Innovative Business Groups (AEI) promoted by the Ministry of Industry, Tourism, and Trade of the Government of Spain, with funds from the Next Generation EU program, through the Recovery, Transformation, and Resilience Plan.
The system developed for the first phase of the project enables real-time monitoring and control of agricultural areas and urban spaces, such as parks and gardens for resource use optimization and to ensure a more sustainable and efficient management of irrigation. A key aspect of the system is the use of 'no-code' techniques, allowing anyone to create applications and websites without programming knowledge, to develop artificial intelligence models. This empowers farmers to create AI models adapted to their crops and specific areasdriving the digitalization process of agricultural companies.
What are the challenges faced by farmers with traditional management of green areas?
Agriculture management is more complex than it appears; it's not just about planting and harvesting. Each crop has its own processes, needs, and analyses. This complexity entails several challenges for farmers, in addition to various external factors that can have both positive and negative consequences for the field. Some common challenges faced by traditional field management include:
The implementation of AI and ML technologies in agricultural systems presents a new challenge for farmers: distrust of the unknown and fear of "losing control." However, this distrust is gradually diminishing over time as the benefits of AI and ML become more apparent.
Each collaborator contributes to the project by leveraging the technologies they excel in and adhering to a predefined plan. In the initial phase of the project, there was a need for an integrative AI platform capable of generating use cases rapidly without requiring coding. Consequently, AI Agro Unifier, built on TuplOS, emerged as the ideal solution. Tupl Agro has brought three pivotal aspects to this project:
At the project's outset, finding an integrative platform for generating use cases with a farmer-friendly interface was paramount. After thorough market analysis and evaluation, AI Agro Unifier emerged as the solution of choice. Built on Tupl's TuplOS platform, AI Agro Unifier offers comprehensive automation capabilities, meeting all project requirements by automating processes and providing a complete service. These requirements included the ability to integrate with various data sources, customize the platform for different use cases, and apply it in an interconnected manner to achieve the final result: intelligent irrigation.TuplOS, our MLOps platform, was designed to enable the creation and customization of use cases according to customer preferences. Its versatile technology, requiring no programming knowledge, makes it ideal for users without coding experience, which was a key pre-requisite in the project's initial phase.
In this phase, various components were developed to capture data from diverse sources, including satellites, weather stations, sensors, and PLCs (for example the Cordis one in this project). AI Agro Unifier plays a crucial role by automating three essential processes: data ingestion, data processing, and data communication.
The development of control processes - such as the Farm Process, Device Process, Climatic Process, and Treatment Process - is another vital functionality offered by TuplOS. These processes enable visualization of integrated data, alerting based on models, and the creation of a general process for executing global calculations with data from all processes.
Furthermore, a machine vision module has been developed through TuplOS for leaf health detection. This module processes and analyzes photos received from farmers via WhatsApp, providing feedback on leaf health.
For example, if the system receives a photo and recognizes that the leaf is damaged and requires treatment, the General Process can alert the user that treatment cannot be performed because the temperature is above 30ºC, which would be counterproductive. This level of integration and advanced analysis ensures that farmers can make informed and precise decisions, thus optimizing the management of their crops.
TuplOS serves as the backbone of the project, simplifying all tasks. Only application logic, knowledge, and understanding of its use are required to leverage its capabilities effectively.
As previously mentioned, the agricultural field entails a high complexity, requiring not only experience but also the ability to correlate data and apply logic. For instance, understanding the unique climatic requirements of each crop and identifying adverse weather conditions enables us to detect, quantify, measure, and take necessary actions. Ultimately, action is paramount, and knowledge serves as the foundation for informed decision-making.
In light of this, a comprehensive study was conducted for this project to delineate the agronomic and vegetative needs of each crop, along with the influence of weather conditions on various phenological phases and water dependency. This thorough analysis enabled the identification of key performance indicators (KPIs), to determine which specific issues come from each sensor, climatic data, and individual farms. By leveraging this data-driven approach, we can implement targeted measures based on precise information, ensuring effective interventions tailored to specific circumstances. This logical framework underpins our approach, guiding our actions to optimize agricultural management practices.
Each use case undergoes a different study, involving intricate analysis and tailored processing. As such, we approach each case individually within our platform.
3.1. Irrigation use case
In the case of irrigation, we have established synchronization with IoT devices such as tensiometers, weather stations, and satellites to procure daily real-time data on crop conditions. This data enables us to generate personalized irrigation recommendations tailored to the specific requirements of each crop. Within our platform, we utilize models to parameterize the data received from these IoT devices and generate alerts based on predefined parameters. These alerts are then communicated to farmers via platforms such as WhatsApp, providing them with real-time updates on irrigation status and offering personalized recommendations.
3.2. Machine Vision Use Case
The AI Agro Unifier platform utilized in this project serves as an all-in-one solution, comparable to an ERP system, as it encompasses the functions of data collection, storage, organization, analysis, and distribution. Notably, it can gather various types of data, including satellite imagery, weather conditions, temperature, and humidity, among others, without reliance on third-party sources.
A specific use case has been developed for this project, which includes the following functionalities:
The development of the AI Agro Unifier system, powered by TuplOS, allows farmers to submit leaf photos via WhatsApp for analysis. Our solution employs machine vision and machine learning techniques to assess the health status of the leaves. This machine vision module, which detects the health of leaves, has the potential to be applied to all use cases. Currently, we are in the phase of recognizing leaf health, and the next step will be to identify the specific pest or disease.
One of the key advantages of this technology is its user-friendly nature and adaptability to diverse use cases.
At Tupl Agro, we integrate this analysis with the phenological phases of the crop. For instance, if a farmer submits a photo in March, indicating the growth phase when flowers appear, we analyze the common pests or diseases affecting flowers during this period. By parameterizing this data, our AI can identify potential causes based on the crop's phenological phase when receiving a photo of an unhealthy leaf.
Furthermore, the "processes" within the AI Agro Unifier control panel consolidate all farm-related information, offering a comprehensive overview of farm operations. These processes serve as a 'master and commander', providing insights into various aspects of farm management.
The user interface of AI Agro Unifier displays unified data, including satellite imagery, weather conditions, and farm-specific information. For instance, users can track the number of alerts generated for each farm. This comprehensive analysis is presented through a dashboard, offering a snapshot of all managed farms' status.
Users can navigate through different views within the interface to explore specific details of each farm or product.
The project offers a wide array of benefits for crop management and agriculture, contributing to enhanced productivity, sustainability, and cost-effectiveness:
1. Optimization of water use:
2. Increased productivity:
3. Cost reduction:
4. Greater efficiency and control:
5. Environmental sustainability:
6. Scalability and adaptability:
7. Access to information and technology:
The second phase of the project, currently under development, aims to automate irrigation processes on farms without the constant supervision of farmers, relying on an AI program. This automation ensures timely action, mitigating the negative consequences of inadequate or delayed irrigation.
Farmers will receive alerts and notifications regarding irrigation needs. Failure to irrigate on time could adversely affect the production of crops such as avocados, mangoes, strawberries, etc. The automation solution guarantees that irrigation needs are addressed promptly, irrespective of the farmer's availability, thereby safeguarding crop health and productivity.
While farmer skepticism persists due to factors like limited knowledge, perceived loss of control, and the absence of tangible examples, ensuring an initial phase with optimal, precise, and tangible results is crucial. Observing the efficiency and effectiveness of AI operations on their farms during this phase can bolster farmer confidence in AI technology. Ultimately, farmers may be more inclined to entrust the entire irrigation automation process to AI.
Call: AEI 2023 Aid
File Number: AEI-010500-2023-207
Project Duration: 1/6/2023-17/4/2024.
Project Budget: 415,335 €.
Granted Subsidy: 315,867 €.
Smart City Cluster
Project coordinators and managers.
INNOVA IRV
They have evaluated various options for terrestrial robotic platforms to best adapt to the avocado cultivation terrain. Additionally, they designed the project's data architecture, enabling efficient storage, management, and analysis of the collected data, including the development of APIs for data access and analysis.
University of Malaga (UMA)
They provide a drone equipped with a multispectral camera and a terrestrial robotic platform with a 3D LIDAR. These hardware devices are capable of obtaining highly precise terrain data. The resolution of the drone's cameras allows capturing images with a ground sampling distance (GSD) of up to 2 cm/pixel, depending on the flight altitude. The LIDAR, on the other hand, captures the environment's structure and enables its reconstruction. Together, they can create a high-precision map of the farm.
Cordis SUITE
They provide a PLC that translates the data collected by the tensiometer, acting as an intermediary between the IoT devices and Tupl. This facilitates the collection of data from all types of tensiometers and enables communication with the irrigation machine in phase two. Cordis's ultimate goal is to be able to connect to any type of device and communication signal to translate the information.
Trops
We are grateful for Trops' contribution to the project, as they supplied valuable crop data gathered by their tensiometers for analysis and insights. This data played a crucial role in developing the models essential for the project. While Trops is no longer actively involved in the project, their collaboration was instrumental in acquiring pertinent data and progressing towards the project's objectives.
The project harnesses artificial intelligence (AI) to enhance the automatic management and smart irrigation of green areas, tackling pressing issues like water scarcity and climate change. Developed by a consortium of organizations, it emphasizes automated irrigation systems and agricultural management through AI and "no-code" techniques, simplifying the process for farmers to create tailored models.
A key component of the project's success is the TuplOS platform, which integrates diverse data sources and automates critical processes such as data ingestion, processing, and communication. Moreover, TuplOS features a machine vision module enabling farmers to assess crop leaf health using images sent via WhatsApp. This initiative aligns with broader objectives to digitize and optimize agricultural practices, enhance water efficiency, boost productivity, cut costs, and foster environmental sustainability.
Looking ahead, the project's second phase aims to achieve full irrigation automation without continuous farmer oversight, aiming to address initial farmer skepticism by demonstrating precise and effective outcomes in the initial phase.
Overall, this project represents a significant stride towards smarter, more efficient, and sustainable agriculture, leveraging advanced technologies to address current and future challenges in the agricultural sector.
