Modular robotic platform for an automated soil monitoring system

Background. The modular robotic platform is implemented using the “robot-constructor” principle. The basic platform includes standardized interfaces for connecting various modules – specialized chassis for different types of surfaces, manipulators for cargo handling, and sensor systems for navigation and environmental monitoring. This architecture allows the robotic platform to be quickly adapted to specific customer needs without the need to develop a completely new solution. The platform demonstrates particular practical value in agro-ecological monitoring, where the modular architecture allows for the rapid adaptation of sensor equipment for analyzing key soil parameters.

Purpose. To present the architectural and functional design of a modular robotic platform implementing the “robot-constructor” principle and to justify its effectiveness as a basis for creating adaptive ground systems within the national aerospace infrastructure.

Materials and methods. The development of a modular self-propelled robotic platform was carried out within the framework of system engineering: conceptual design → synthesis of architecture → selection of components → integration of subsystems → verification on a physical layout. The design is based on a lightweight and rigid metal frame that allows quick replacement of modules (chassis, manipulators, sensors). For work in the agricultural sector, it is possible to switch from a wheeled to a tracked base.

Localization and orientation are implemented using visual odometry and simplified SLAM (ORB-SLAM2 light) for building 2D maps. Motion control is a multi-contour PID controller: the external contour corrects the deviation from the trajectory according to the video (P+D), the internal one stabilizes the speed according to the encoder data (I-component). The software platform is ROS
2 Humble (Python 3.10). Key nodes: - vision_node – marker recognition
(OpenCV + TensorFlow Lite); - navigation_node – route construction and correction (RRT); - control_node – engine control with adaptive PID adjustment depending on the weight of the cargo; - telemetry_node – data export to JSON/CSV and integration with ERP/MES via REST API. A sensor module is used to monitor the soil: multispectral cameras, humidity, temperature, pH, nutrient sensors, and a sampling device.

Results and conclusion. During the project, a modular self-propelled robotic platform was developed and physically prototyped, functioning as a universal ground component within the domestic Aeronet ecosystem. A unified mechanical and electrical platform with standardized connection interfaces (mechanical – quick-release dovetail mounts; electrical – GX16-4P industrial connectors; software – ROS 2-compatible topics), ensuring the modularity of the chassis, manipulators, sensor complexes, and actuators.

A visual navigation system has been developed and tested.

Based on OpenCV and fine-tuned YOLOv5, an algorithm for recognizing color lines, QR codes, and natural landmarks has been implemented. The platform has been integrated into the educational process at Melitopol State University in four areas of training.

The study confirmed the fundamental feasibility and high efficiency of the modular robotic platform as a tool for converging the Aeronet and Technet NTI roadmaps. The developed solution successfully combines the characteristics of technological sovereignty (domestic component base, open-source stack, rejection of dependent technologies), economic affordability, and functional flexibility.

The practical significance of the project is due to its dual purpose:

1) as an import-substituting industrial solution for the automation of intra-plant logistics at small and medium-sized enterprises;

2) as a multifunctional educational and research platform that forms a personnel reserve in the field of robotics, AI, and digital manufacturing.