Specialist system for diagnosis and prediction of infestations and failures in sugarcane crop areas

• Grant number: 19/15567-2
• Support Opportunities: Research Grants - Research Program in eScience and Data Science - PIPE
• Start date: August 01, 2020
• End date: October 31, 2022
• Field of knowledge: Physical Sciences and Mathematics - Geosciences - Physical Geography
• Principal Investigator: Marcelo Elias Delaneze
• Grantee: Marcelo Elias Delaneze
• Company: IDGeo Inteligência em Dados Geográficos Ltda
• CNAE: Cultivo de cana-de-açúcar ; Atividades de apoio à agricultura ; Atividades profissionais, científicas e técnicas não especificadas anteriormente
• City: Piracicaba
• Pesquisadores principais: Luiz Henrique Pereira ; Ronan José Campos
• Associated researchers: Cláudia Maria de Almeida ; Roberta Averna Valente B Tolini
• Associated research grant: 17/22870-8 - Diagnosing sugarcane crop problems through satellite images, AP.PIPE
• Associated scholarship(s): 21/00389-1 - Development of an expert system for diagnosis and prediction of agricultural productivity, BP.TT; 20/15541-0 - Evaluation of spatial models for diagnosis and prognosis of problems in sugarcane fields, considering distinct agro-environmental conditions and different scenarios, BP.TT; 20/15534-4 - Field monitoring for diagnosis of low productivity stains in sugarcane - Focus on areas of partner plants, BP.TT; 20/09035-5 - Specialist system for diagnosis and prediction of infestations and failures in sugarcane crop areas, BP.eScience.PIPE

Abstract

There is a growing concern among sugarcane producers to improve the productivity and reduce losses due to infestations of pests, diseases, weeds and failures derived from mechanized operations, which directly affects the development of the crop. The current methods for identifying areas of low productivity are based mainly on images collected by UAV (Unmanned Aerial Vehicles). However, this approach has technical limitations due to the large volume of data collected, the need for a robust infrastructure, processing time, lengthy time between acquisition and results, high operating cost and lack of qualified staff. In this context, this research proposes to develop methodological procedures based on machine learning, to establish the association between the manifestation of the problem (dynamic behavior of low productivity stains, and context elements) with its cause (diagnosis). This, integrating the "multisensor orbital" approache and cellular automata to model and predict the evolution, and/or manifestation, of problems in sugarcane fields throughout the harvest; considering the influence of the main physical, meteorological and management conditions that contribute to its occurrence. Furthermore, it also aims to develop a tool for systematic monitoring of crop problems, based on the integration of the prognosis process with the diagnosis, in order to promote agility in identifying the causes, estimating the losses, and directing corrective actions. Finally, the achieved results will be displayed into an online platform in the format of reports and interactive management panels with the presentation of the information in different levels of context. (AU)