Embedded Files
Note: These models could be further optimized with a larger dataset, although this requires considerable time, technical expertise, and computational resources. The same applies to autonomous robotic systems, which integrate these models with actuators capable of performing heavy and prolonged tasks. Therefore, I encourage business leaders and investors to consider reaching out to research centers or qualified technicians to propose projects and invest in advancing technological innovation in our beautiful country, Ecuador.
Note: These models could be further optimized with a larger dataset, although this requires considerable time, technical expertise, and computational resources. The same applies to autonomous robotic systems, which integrate these models with actuators capable of performing heavy and prolonged tasks. Therefore, I encourage business leaders and investors to consider reaching out to research centers or qualified technicians to propose projects and invest in advancing technological innovation in our beautiful country, Ecuador.
Research Lead at AIRIX (October 2024 - Present)
Research Lead at AIRIX (October 2024 - Present)
I lead R&D at CacaoHealth Detector project of AIRIX, a precision agriculture initiative backed by UNESCO and Nestlé. I manage a team of 8 engineers in developing a functional autonomous rover (Computer Vision, ROS Humble, ArduPilot) for early disease detection in cocoa crops.
As an Edge AI specialist, I design embedded electronic systems and implement computer vision models (YOLO, RF-DETR) quantized for Jetson Orin devices. I optimize our training cycles through auto-labeling pipelines using SAM and synthetic data generation. Beyond technical leadership, I manage the project's financial and strategic resources, ensuring scalability and social impact within the agricultural sector.
Results & Impact
Results & Impact
To learn more about our methodology and milestones, contact us at marketing_ec_01@airixtech.com or visit airixtech.com | (1) Airix tech - YouTube . AIRIX operates as an autonomous research unit. The technological development and intellectual property (IP) of the project belong exclusively to its active partners, with strategic support from international organizations such as UNESCO and Nestlé.
Team and Roles
Jefferson Ramirez – Project Lead & Computer Vision Lead and Electronics
Leads the project, coordinates the team, and oversees AI development, system integration, and technical validation. Represents the project and manages funding.
Juan Saeteros – Electronics and Navigation Lead
Responsible for prototyping and simulation of electronic and navigation systems. Coordinates integration across areas and can assume leadership in the director’s absence.
Erick Mendoza – Mechanical Lead
Responsible for the design, construction, and maintenance of the robot’s mechanical system, as well as component manufacturing.
Joshua Cobos – Navigation, Vision, and Electronics
Supports the development of autonomous navigation, vision, and hardware-software integration. Participates in simulation and field testing.
Team and Roles
Milena Rodríguez – Mechanical
Supports mechanical design and assembly, quality control, and system validation during testing.
Rolando Mendieta – Electronics and Navigation
Supports development and integration of electronic and control systems, troubleshooting, and robot testing.
George Yaguana – Mobile App and Web Lead
Develops the mobile application and communication services with the robot, ensuring connectivity and system integration.
Diego Salazar – Marketing and Machine Learning
Responsible for commercial strategy, market analysis, and support in machine learning models and investor presentations.
Research student (May 2024 - September 2025)
Research student (May 2024 - September 2025)
My activities will involve configuring and calibrating a JetBot robot, designing and executing mapping routes to cover the area of the corn and cacao field, collecting field data by operating the robot, and processing that information to generate 3D maps of the crop.
The activity we are going to carry out resembles the one shown in the image from the University of Sydney, specifically from the Australian Centre for Robotics with its robot 'RIPPA'.
Robotics
Robotics
The research focuses on developing an ultra-compact, low-cost autonomous robot for the automation of maize and cocoa phenotyping, utilizing computer vision and artificial intelligence. This robot will be equipped with GPS, LIDAR, GPU-powered microcomputers, 3D cameras, and SSD algorithms and will use GPS-based navigation methods and waypoints to move through agricultural environments. Its primary objective is to collect key data on crop growth, pest detection, and resource optimization. This research aims to enhance agricultural sustainability by reducing costs and minimizing pesticide use, aligning with the SDGs related to zero hunger, efficient water use, responsible production, and decent work.
Computer Vision Models
Computer Vision Models
These computer vision models were trained by me to demonstrate their application in various fields, such as industrial safety, agriculture, and public safety. Additionally, the last video shows the deployment of an open-source model, selected to test the functionality of these models on an embedded system, in this case, the Jetson Orin Nano Developer Kit.
inferenciaArandanos.mp4It is important to highlight that these models represent a significant advancement towards system automation and must operate in conjunction with sensors and actuators, depending on the application. In the field of industrial safety, for example, the model could alert the plant supervisor when workers are not complying with safety codes. In the case of the blueberry detection model in agriculture, it could be used to facilitate precision farming by counting blueberries ready for harvest and those that still need to ripen. Finally, the model for public safety could recognize high-risk situations and autonomously alert authorities or security companies, enabling real-time actions to prevent incidents.
Page updated
Google Sites
Report abuse