Harvest helper: Autonomous vehicle offloads grain on the go
Total farm production nearly tripled in the United States from 1948 to 2017, according to the U.S. Department of Agriculture. We’re aiming to keep that arrow pointed upward, especially as per capita agricultural land continues to shrink and labor shortages challenge agricultural producers. We believe more autonomous systems are a big part of the solution, and our research most recently has focused on pairing an autonomously tractor-pulled cart with a combine harvester to automatically offload the grain.
Harvest is one of the most stressful times of the year for farmers. They operate in tight weather windows, racing to get crops like corn and soybeans out of the field before significant grain loss occurs. Unloading on the go is key to a high-efficiency operation — depending on yields, stopping to unload each batch of harvested grain can reduce efficiency by about 30 percent.
Unloading on the go is a complex task. Currently, skilled tractor operators must continuously communicate with one another to manually synchronize the harvester and grain cart in order to maintain positional accuracy between the two vehicles and monitor fill levels. Working outdoors in changeable and invariably less-than-perfect conditions, the operators have to weigh the risk of spilling crop against losing efficiency.
A combine operator has to execute these tasks while keeping tabs on obstacles and terrain changes in the field. Crop conditions also can vary throughout the field, requiring the operator to adjust the harvester settings to maximize productivity. A system that autonomously controls the grain cart during unloading would lower operator stress and fatigue and enable more efficient operation, maximizing grain transfer and continuity of the harvest operation.
A Purdue research project in which I participated was one of the first steps in proving the technology. Our system leveraged John Deere’s existing Machine Sync system (which guided the approaching grain cart tractor into position for unloading) and machine vision (which created a digital image of the grain) to measure the amount of grain in the cart and autonomously move the cart during unloading. Our automatic offloading controller took the input from the machine vision system and determined when and which way the cart needed to move to complete the fill. Machine Sync handled process steps like sharing the harvester’s location, lining up the grain cart, and executing the relative position changes commanded by our automatic offloading controller.
The Purdue work was funded by John Deere, with Greg Shaver, a professor of mechanical engineering, as principal investigator. Also on the project team, besides myself, were Daniel DeLaurentis, a professor of aeronautics and astronautics, and Tony Vyn, the Henry A. Wallace Chair in Crop Sciences and professor of agronomy in Purdue’s College of Agriculture.
Some of the biggest research challenges revolved around operating conditions. Trying to use machine vision systems in dusty environments is extremely demanding. Another difficult aspect was devising the control algorithms for large machines that have to be perfectly synchronized for the system to function properly. Our research and development work has led to the submission of one patent application, in tandem with Deere, around the auto-offloading system. We are transitioning our work to Deere, which will take the next steps toward commercializing the technology.
As our research helped demonstrate, the best role for automation in agriculture’s future is to help offset labor shortages, improve efficiency, and enable more data-driven decision making — all of which will help move us toward a more environmentally and economically sustainable system.
John T. Evans IV, PhD
Assistant Professor, Department of Agricultural and Biological Engineering, and Department of Mechanical Engineering (by courtesy)
Faculty Council Member, Purdue Engineering Initiative in Autonomous and Connected Systems (ACS)
Faculty Team Member, The Open Ag Technology and Systems (OATS) Center
College of Engineering