Robotics (both ground and aerial) and machine learning (e.g. deep learning) are expected to dramatically change our work and life. Plant sciences and in particular agriculture is one of the most important fields where the two technologies would have a significant impact. Agricultural robots can assist (or replace) humans to work in harsh field conditions and regions with limited labor. We are developing custom robots and robotic networks with machine learning capabilities for various agricultural tasks such as phenotyping, production management (e.g. weeding and pruning), and harvesting. With the advent of the big data era, machine learning techniques will help transform the way we observe and understand plants and crops. Our lab has developed a technique to use images from the unmanned aerial systems and convolutional neural networks to count cotton flowers.
Xu, R., C. Li, A.H. Paterson, Y. Jiang, S. Sun, J. Roberson. Cotton bloom detection using aerial images and convolutional neural network. Frontiers in Plant Sciences. doi: 10.3389/fpls.2017.02235.
Xu, R., C. Li, A.H. Paterson, Y. Jiang, S. Sun, J. Roberson. Multispectral imaging and unmanned aerial systems for cotton plant phenotyping. PloS one.