As part of a small team of multi-disciplinary researchers, we are developing a ground-based mobile robot used to autonomously survey and phenotype crops for plant geneticists and farmers. Our robot, known as "The Robotanist", utilizes a series of non-contact sensors (LIDARs, stereo cameras) and contact sensors (spectrometers, penetrometers) to extract relevant biochemical and physiological characteristics. In short, the goal of this project is to automate, accelerate, and add reliability to the process of crop monitoring and phenotypic data collection of plants.
My research is centered around robotic grasping of leaves in the field (mainly of maize and sorghum). This process includes the real-time 3D reconstruction and modelling of crops in order to detect leaves from live stereo images. The second half of my research explores using kinematic and manipulation techniques to control our robotic arm to grasp a detected leaf. Once a leaf is grasped, a spectrometer is applied to acquire a unique spectrum of reflected light. These representative spectra are then used to train learning algorithms for predicting compositional traits of the plants.
