Implementación de un prototipo de robot teleoperado para detectar irregularidades foliares en cultivos de fresa usando Visión Artificial.

Abstract

The objective of the present titration work was to implement a teleoperated robot prototype to detect leaf irregularities in strawberry crops using artificial vision. The mechanical design was made in SolidWorks; for the control of the motors Arduino was used and for the artificial vision, the programming language Python. The prototype called FresBot is able to travel irregular terrains, thanks to its traction system, it can be mobilized autonomously and by tele operation. On the other hand, the system provides a view of the detection process on a Raspberry touch screen that has been incorporated and generates a report in a txt file with the results obtained. The field tests of FresBot prototype were carried out in the "La Florida" neighborhood, belonging to the Cevallos canton of the province of Tungurahua, with a sample size of 100 rows with 100 strawberry plants each, with an estimated age of 4 months. The results obtained after evaluating the FresBot prototype, showed that the certainty of connection of the radiocommunication control (RC) with its receiver module is only 65%; with regard to autonomous operation, 80% of successful tests and 20% of failed tests were obtained, 75% of which were failures due to the robot overturning due to the irregularity of the terrain; in terms of repeatability, the prototype does not show any variation between tests, and finally, by means of a visual inspection of the number of plants with irregularities identified by the Agronomist vs the FresBot prototype, the 97.53% effectiveness of the system was obtained. The observed results express that the foliar irregularity detection system works correctly, but, a communication protocol different from the applied one should be used, in order to improve and / or avoid communication failures, therefore, it is recommended to implement it in future works.