Automatic pest traps and geostatistics applied to integrated pest management

• Grant number: 17/08195-6
• Support type: Research Grants - Innovative Research in Small Business - PIPE
• Duration: December 01, 2017 - March 31, 2020
• Field of knowledge: Agronomical Sciences - Agronomy
• Cooperation agreement: FINEP - PIPE/PAPPE Grant
• Principal Investigator: Marcus Vinicius Sato
• Grantee: Marcus Vinicius Sato
• Company: Agrosmart S/A
• City: Campinas
• Associated scholarship(s): 18/16066-4 - Automatic pest traps and geostatistics applied to integrated pest management, BP.TT

Abstract

Pests are one of the main factors that limit agricultural yield. Annually, Brazilian agriculture loses about US$ 14.73 billion and 7% of agricultural production due to the pest attacks on the crop. In Brazil the most affected crops are sugarcane, corn and soybeans. And the main pests affecting these crops are the armyworm (Spodoptera frugiperda), the Helicoverpa armigera, and the populations of Chrysodeixis includens (Lepidoptera: Noctuidae). To reduce the losses caused by these pests, it is necessary to perform an agile and efficient pest management and monitoring. However, the current monitoring method is laborious, time-consuming, expensive, and sometimes prone to errors, which prevents farmers from achieving performance and cost goals. Therefore, the objective of this project is to create a system to monitor agricultural pests from a distributed imaging device operated by a wireless sensor network that is capable of automatically acquiring and transmitting images from the capture area (Traps) to a remote control platform, which will allow access to data over the internet. To achieve this goal will be used pheromone traps that attract the insect and trap it in adhesives composed of entomological glue. These pheromone traps will also contain an embedded camera to obtaining captured insect images, and a module for sending approved images to the cloud. With the images in the cloud will be applied a classifier algorithm of images; By the methods of support vector machines, artificial neural networks, k-nearest neighbors; Which will be able to characterize the morphometric measurements of the insects and consequently to identify them. With a network of traps spread in an area it will be possible to identify the exact moment of the intervention with chemical or biological control, as well as the best defensive to be applied, it will also be possible to quantify the level of infestation, the places in need of chemical control and finally to correlate the incidence of pests with other data available in the monitored area. With this information, farmers will be able to obtain a more efficient and agile pest management, improving production quality, reducing the environmental impact caused by the agricultural activity, saving on the cost of inputs and increasing the efficiency of the workforce in the field. (AU)