Developing Sensing and Robotics Technologies for Plant Phenomics

Abstract

Feeding a world population of 9 billion people by 2050 amidst increasing climate variability is one of the greatest challenges facing humanity. The genomics revolution provides unprecedented power to develop new and advanced crop cultivars with the gene combinations needed to address these global challenges. However, rapid and repeatable measurement of crop phenotypic parameters remains a major bottleneck in plant genomics and breeding programs. High-throughput phenotyping technologies that can quickly and repeatedly scan tens of thousands of individuals using an array of advanced sensor and data analytics tools are critical to improving the ability of scientists to dissect the genetics of quantitative traits such as yield and stress tolerance. Our interdisciplinary team is developing a robot-assisted field-based high throughput phenotyping system that integrates both ground and unmanned aerial elements to quantitatively measure a suite of key traits iteratively throughout the growing season. This project is expected to unmask plant responses that will inform a new level and quality of decision-making in the selection of crop genotypes for specific production conditions. A deep learning convolutional neural network was used to identify flowers and the flowering peak of cotton plants. The task coordination between ground and aerial vehicles will result in new discoveries in the area of partitioning and coverage control. In a different but related project, the talk will introduce a Berry Impact Recording Device (BIRD) sensor that has been successfully developed to simulate a real berry to quantitatively measure mechanical impacts on the real fruit created by machine harvesters, packing lines, and transportation vehicles.