Robotic gantry for automated imaging, sensing, crop input application, and high-throughput analysis

Greenhouse-level high-throughput analysis for automated plant imaging and interaction is important for the identification of productive and resource efficient genetic crop lines. This paper discusses the development of an automated gantry system that allows specification of and interaction with individual plants in a 64-square-meter (900-square-foot) greenhouse plot. The system minimizes custom hardware to allow construction using mostly Commercial-off-the-shelf (COTS) components so that greater resources could be allotted for imaging, 3D plant mapping, and fertilizer application systems. A further improvement on existing commercial systems which is explored is an Internet-of-Things (IoT) capability with machine learning programs in the cloud. This capability allows the system to both automatically report findings of scheduled greenhouse plot checks and also respond appropriately when necessary to anomalies; taking images, for example, or additional fertilizer spraying. This system allows highly-customizable imaging and plant interaction capabilities throughout the growing period of the plants enabling exploration of previously inaccessible information about crop reactions to growing conditions. The system also makes this information immediately available to researchers who aren't physically present in the greenhouse, improving efficiency and expanding opportunities for collaboration.