Forest Management 4.0: Integrated Tree-Level Forest Inventory Methodology

• Grant number: 20/05900-3
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Start date: December 01, 2020
• End date: August 31, 2021
• Field of knowledge: Interdisciplinary Subjects
• Principal Investigator: Maycow Lucas Dutra Gomes Berbert
• Grantee: Maycow Lucas Dutra Gomes Berbert
• Company: Treevia Serviços de Engenharia Ltda. - ME
• CNAE: Atividades de apoio à produção florestal ; Fabricação de aparelhos e equipamentos de medida, teste e controle ; Desenvolvimento e licenciamento de programas de computador customizáveis
• City: São José dos Campos
• Associated researchers: Emily Tsiemi Shinzato ; Esthevan Augusto Goes Gasparoto ; Marcelo Handro Cordaro
• Associated scholarship(s): 21/03940-0 - Development of the Android mobile application for interaction of IoT end-devices with the cloud layer, BP.TT; 21/01137-6 - Development of the radio frequency (RF) layer that makes up the "SmartForest: Single Tree" System., BP.TT; 21/00533-5 - Development of geoprocessing routines and application of new biometric algorithms based on AI (Artificial Intelligence) to support the single trees inventory paradigm, BP.TT; 20/16219-5 - Development and implementation in cloud-computing of forest algorithms for processing inventories at the individual tree level: Forest Management 4.0, BP.PIPE

Abstract

Digital technologies are revolutionizing industries around the world and in different sectors of the economy, from the manufacturing to health sectors. In agriculture is no different, accurate and automatic harvesters are already reality. The forestry sector is traditionally known to have rather long innovation cycles; however this scenario may change. Field data collection through the allocation of sample plots of fixed or variable area is the standard methodology that has been applied in forest inventories since the beginning of forest science. Such information is always associated with a unit of equivalent area that is the area of the sample plot, and then extrapolated to the known area of the forest stand of interest. However, with the current technological advances observed in cartographic and remote sensing science (satellites and drones generating high resolution images), accurate counting of the number of trees that make up a forest stand is a challenge that can be overcome through the application of robust algorithms for remote sensing, computer vision and machine-learning. With this great technological leap, there are still great opportunities for paradigm shifting within the context of forest monitoring and measurement. Treevia's forest management solution 4.0 is not only part of this change, but is also one of the pioneering solutions in the world by bringing the Internet of Things into the context of monitoring planted forests. The Forest Management 4.0 solution involves an Integrated Forest Monitoring System "SmartForest: Single Trees", a joint software and hardware solution that enables the automation of forest inventories through an innovative concept known as individual tree sampling. The objective of this proposal is to study the technical feasibility of a precision forest inventory methodology, using as one of the innovations the use of individual trees in the sampling process, unlike current methodologies that use forest plots (tree groups) as sample units. Once proven viable, the solution will be transformed into a new product in the startup portfolio, the "SmartForest: Single Trees". In order to do so, intermediate and feasible solutions of the methodology will also be adapted and developed for the development of IoT hardware and sensors, software and firmware development, application and consolidations of advanced mapping, segmentation and mapping techniques, and mainly application and algorithms within forest science. It is hoped that the results obtained will enable a complete change in current forest inventory methodologies. The methodology will allow a reduction of up to 5x the number of trees measured, allow for remote forest monitoring from anywhere in the world and open up a range of possibilities in forest science through the massive collection of data at high temporal frequency, all with an expected potential cost reduction of up to 30% compared to current methodologies. (AU)