Contributions to the configuration of fleets of robots for precision agriculture

Zusammenfassung

Precision Agriculture (PA) aims to the application of selective treatments as well as the use of agricultural inputs depending on the field and crop needs spatially localized. One of the most demanded agricultural work is the effective weed control, to increase productivity while decreasing the usage of polluting chemical products. The incorporation of new positioning technologies (GPS, Laser), acquisition and detection (computer vision) and automatic control adapted for vehicles (tractors) and for agricultural machinery (implements), has established the foundation for the development of PA. Although, only a few cases have achieved a configuration of a fully autonomous agricultural system, and with a minor dimension, a configuration of a fleet of robots for cooperation and for distribution of agricultural tasks. The study presented in this research includes three major contributions oriented to the aforementioned automation. More specifically, this research provides the foundation through the development of a control architecture for the integration of both an autonomous vehicle and an autonomous implement, endowed with the technology required to achieve such autonomy and efficient performance. The first contribution is the development of a simulation environment to study and evaluate the implementation of PA techniques that allows the performance, cooperation and interaction of a group of autonomous robots to be analyzed, while the execution of a specific task is simulated in a three-dimensional world. The second contribution is the development of a proposal of a control architecture to integrate a vehicle equipped with an implement, with the purpose to constitute a fully autonomous agricultural unit able to work cooperatively in a fleet of robots. The third contribution is the integration and validation, in a real crop field, of a perception system (based on computer vision and GPS location), an actuation system (consisting of an autonomous tractor and an autonomous implement for mechanical weed control) and a decision-making system (which is the core of the integration of these elements), all encompassed in the control architecture developed from the perspective of full autonomy. This integration was performed using the selection, management, configuration and synchronization of these systems, providing a model for a fully autonomous vehicle for agricultural applications. Successful results of several experiments conducted on real crop fields in differentseasons and under different crop conditions are presented, demonstrating the performance and validity of the integrated proposal in guidance and weed control tasks in a maize field, and its usefulness and effectiveness. This work is a major advance in the design of mobile units able to work in coordination as a fleet of robots in agricultural tasks, resulting in several publications in various scientific journals and prestigious international conferences in the area of PA. The developments obtained have been a fundamental part in the progress of the RHEA project, demonstrating the ability to configure a fleet of robots for PA applications, having been endorsed by members of the project consortium, as well as positive evaluations of the representatives of the European Union, both technical and management.