The Grain Quality Monitor is a vision and learning-based approach for detecting and estimating the composition of grain flowing through a combine harvester. It is implemented as a module in John Deere’s “Auto Maintain” function, which uses this information to perform closed-loop control of machine settings to maintain performance goals. This product was awarded an innovation “Silver Medal” at AGRITECHNICA 2015.
Previously, to maximize effectiveness, an operator had to manually configure and reconfigure the combine’s settings for important factors, including:
When field conditions change, which occurs regularly throughout a daily harvesting operation, machine settings are often no longer optimal.
Now, with our Grain Quality Monitoring system and the combine harvester, John Deere’s “Auto Maintain” updates these settings automatically, maintaining optimal performance. In a typical farming operation of several thousand acres, these benefits can represent savings as high as hundreds of dollars per hour.
A custom-designed camera images the grain as it moves through the harvester. A machine- learning pipeline extracts and processes color, texture, and segment features to determine the composition of the material for the various classes of interest. Our system applies a calibration function to produce percent weight composition estimates over time. The system combines grain quality data with information from the combine harvester to determine appropriate closed-loop control actions.
Video feeds of the grain moving through the machine and visualization of the system’s detections are available to the operator on the in-cab display.
The system is designed and tuned to be robust to a wide variety of appearance variation in grain across five supported crops: corn, soybean, rapeseed, wheat, and barley.
The Image Processing Module provides the embedded computational resources required to run the application, with a ruggedized design ensuring continuous operation in challenging agricultural environments.
Following NREC’s tradition of solid engineering, Grain Quality is based on a reliable hardware platform.
National Robotics Engineering Center
10 40th Street
Pittsburgh, PA 15201
+1 (412) 681-6900
Carnegie Mellon University
Legal Info | www.cmu.edu