March 01, 2013      

Specialty crop farmers toil from dawn to dusk in their fruit orchards, but they may soon have robotic farming companions to make that work a 24/7 proposition.

Productivity equals profitability

“Once the technology becomes fairly turnkey and well-integrated and growers realize the productivity and profitability, we will see that robots really have an advantage over humans in some tasks,” said Daniel Schmoldt, program leader at the U.S. Department of Agriculture’s National Institute of Food and Agriculture.

One natural area would be in overnight crop monitoring and spraying as robots can provide their own lighting, Schmoldt said. “Humans in the fields generally don’t do well after daylight,” he said.

Schmoldt’s office oversees two separate projects at Carnegie Mellon University’s Robotics Institute in Pittsburgh that are developing automated farming systems for specialty crops, one focusing on apples and the other on oranges. The combined annual crop yield for both fruits is nearly $5 billion.

The projects are called the Comprehensive Automation for Specialty Crops (CASC) program and the Integrated Automation for Sustainable Specialty Crop Farming Project, respectively. The former received a four-year, $6 million grant and the latter a three-year, $4 million grant.

Funding for both was through the USDA’s Specialty Crop Research Initiative, established as part of the 2008 Farm Bill to solve critical issues facing specialty crops, which includes fruits and vegetables.

Both grants were matched dollar for dollar by industry, state government and other sources. Although funding through the Specialty Crop Research Initiative has now expired, Schmoldt said that USDA support continues through other funding, including the National Robotics Initiative.

Driverless vehicles central to multiple solutions

The robotic farming projects are looking at a number of areas, including autonomous vehicles, digital pest traps, electronic tree counters, and new designs for mechanical harvesting. The emphasis is on aiding farm workers rather than replacing them.

The driverless vehicles are at the heart of both projects. Laser sensors and software are mounted on the robotic vehicles and on platforms they pull that rove throughout the orchard to gather data on tree health and crop status.

The vehicles can also administer precise amounts of water or chemicals to specific areas or trees, in addition to automating such routine tasks as mowing.

Grafting robotics to agricultural needs

“Our project has two main thrusts: crop intelligence and labor intelligence,” said Sanjiv Singh, CASC project director. “On crop intelligence, we want to provide growers with timely and high-resolution data on such things as disease, insect infestation, growth rate and prospective yield.”

Toro from CMU

CMU’s modified Toro

Singh explained that any kind of manufacturing process starts with good information. “Typically, farmers have not had this data. If you don’t know what the problem is, you have no chance of mitigating it,” he said.

“What we are doing is grafting state-of-the-art robotics to meet the needs of agriculture,” said Singh. “We are educating the growers as to what is tactically feasible and technologically viable.”

For example, knowing how much fruit is on the trees is a big issue for growers. “A bumper crop can be as disastrous as a bad yield,’ he said. “If the cost of the harvest is greater than the contract value, the farmer loses money.”

On the labor side, Singh said that helping workers be more efficient is a priority. “Specialty agriculture is very labor-intensive,” he said. “We need to rethink how humans and machines can work together and reduce grower vulnerability to labor availability.”

Presently, farm workers must constantly go up and down ladders to maintain and pick fruit. By automating that and allowing the pickers to remain on the platforms for such tasks as pruning, thinning and harvesting, they can be more productive.

Toro’s crop monitor

One such aspect of the CASC project was converting a two-seat Toro four-wheel drive utility vehicle into an autonomous electric system that monitors crops for fruit health and bugs through lasers that have a 200-degree field of vision.

“It is not a matter of ‘if’ but ‘when’ this product is ready for market,” said Dana Lonn, director of Toro’s Center for Advanced Turf Technology, one of CASC’s industrial partners. “We are working on the value proposition of making the economics work so that farmers can get the job done at less money.”

Lonn added that autonomous vehicles are “not quite good enough” to replace workers and there will still need to be some human interaction. One of the challenges remains making such a vehicle a cost benefit for the grower.

As for a price tag, Lonn estimates that the autonomous vehicle would be roughly twice that of a non-autonomous model that goes for $30,000.

Separately, the Integrated Automation for Sustainable Specialty Crop Farming Project is being run through the Robotics Institute’s National Robotics Engineering Center (NREC).

Three keys: autonomous vehicles, sensing and computing

CMU's modified Deere tractor

Autonomous John Deere from CMU

“As the project started, we looked at a lot of different types of automation. What bubbled to the top as most promising were autonomous vehicles and advanced sensing and computing technology,” said Carl Wellington, senior NREC commercialization specialist.

NREC has developed a GPS-free Navigation system, using visual odometry and an inertial measurement unit, to allow the autonomous vehicles to operate when GPS is lost due to the tree canopy. It can operate standalone or as a supplement to GPS.

Wellington said that his project conducted orange grower focus groups and determined that spraying was a major issue both for controlling diseases and containing costs.

To that end, NREC is developing tree-level precision spraying that utilizes the sensing and data collection capabilities of the autonomous tractors for 3D modeling that accounts for tree height and density.

Each nozzle on the chemical tank can be autonomously controlled to spray only when directed toward foliage, said Wellington, making it both more cost-efficient and safer for the field workers.

“The advanced sensing is not dependent on the autonomous tractors,” Wellington said. “We have tested the technology on manned vehicles to show the value in having sensors and lasers.”

John Deere and grove management

As for the autonomous vehicles, NREC and John Deere modified model 6430 premium tractors to deploy and test a fleet of networked, unmanned vehicles in the orange groves of Southern Garden Citrus, one of Florida’s largest growers.

The tractors perform such tasks as spraying and mowing, while being monitored remotely by supervisors in a control booth for any anomalies that may occur.

Deere is currently investigating the optimum tractor-to-supervisor ratio. “One supervisor can control any number of autonomous tractors, but there are other factors involved, including how complicated and well managed the grove environment is,” said Stewart Moorehead, manager of robotic systems for the John Deere Technology Innovation Center.

“If a tractor spots something, it stops and radios back to the supervisor to await human interpretation,” Moorehead said. “Robots are very patient. If there are too many tractors in the system, they just sit and wait for the supervisor to get to them. There are diminishing productivity gains if it is too high of a ratio.”

Moorehead said Deere selected the tractor model it did because that is what the grower currently uses, but the model is not as important as the technology, which can work on any tractor size.

He could not give a timetable as to when an autonomous tractor might be commercially available.

“It is a long process because there is a lot more than simply technology,” Moorehead said. “There is a chain of activities that needs to get done, including determining how to service and support a vehicle with laser scanning and stereo camera technology. It is going to be an interesting project to commercialize.”

Beyond apples and oranges

Although the two CMU projects focused on apples and oranges, the robotic farming technology being developed can be applied to a variety of specialty crops, including peaches, apricots, pears and grapes, among others.

“These projects are not crop-specific. Long-term, they will have a huge impact on any crop,” said Schmoldt.

The CASC program initiatives were field tested and refined with apple growers in Pennsylvania, Oregon, and Washington and had participation from Penn State University, Washington State University, Oregon State University, and Purdue University, as well as the USDA Agriculture Research Service.

NREC project participants included John Deere, Southern Garden Citrus, the University of Florida and Cornell University.

And when might we see robotics fully implemented in America’s fields and orchards? No one can say with certainty but, as Schmoldt said, “Five years ago, who would have thought we would be talking about self-driving cars on highways today.”