On Campus: Apples and Oranges

You don't expect to find Carnegie Mellon computer scientists working in apple orchards and orange groves. But maybe you should, says Sanjiv Singh, a research professor in the Robotics Institute. "I am very interested in automation in agriculture," he says. "I have gravitated to farm applications because I can see a benefit for society."

Last November, the U.S. Department of Agriculture awarded $10 million to two research groups at Carnegie Mellon to develop systems to monitor crop health, spray water and pesticides, and perform certain automated tasks. One team headed by Singh received $6 million to develop equipment for apple orchards, while another directed by Tony Stentz and Herman Herman of the Robotics Institute's National Robotics Engineering Center received $4 million to create technology for use in orange groves.

The award is part of a USDA grant program called the Specialty Crop Research Initiative, which encourages researchers to design systems to improve the growing of plants that are valuable but difficult to cultivate, including fruits, vegetables and nuts. According to the USDA, research into technology used in orchards and groves hasn't kept pace with advances in the cultivation of bulkier cash crops such as wheat or cotton.

Some people might wonder why researchers from Carnegie Mellon--an urban university not typically associated with agriculture--are working on farming technology. But in fact, the Field Robotics Center has already helped design robots to harvest so-called broad-land crops like alfalfa, and researchers have worked on other projects with agricultural scientists from Penn State, Purdue, Oregon State and Washington State.

 "We're not plant scientists--that's not what we do," Singh says. "But we do have people here who are interested in making an impact in agriculture. Robotics is an interdisciplinary science and agriculture is an area of application."

Many people assume that farm robots would be useful picking apples and oranges from the boughs of trees. But Singh says that humans are better at picking produce. Instead, robots can help with things like orchard management. Fruit growers, for instance, don't always have accurate ways to predict how many apples or oranges they'll have until the growing season ends, but autonomous farm vehicles could drive through orchards and groves and count the ripening fruits.

This past fall, Singh and his colleagues rode an all-terrain vehicle packed with electronic gear around the grounds of Soergel's Orchards in Franklin Park, Pa., north of Pittsburgh. As the ATV chugged through rows of apple trees, researchers collected a range of data about the terrain and land features. The data will help them develop algorithms to control autonomous guidance and obstacle avoidance systems for use in orchards.

Singh and his team hope to create sensors and GPS-guided navigation systems that will enable robots to independently monitor orchards like Soergel's. Some of their sensors might be able to detect chlorophyll levels--a sign of plant health--and could alert growers to potential problems such as diseases or bug infestations.

 "Big producers care a lot about the crop they'll get," Singh says. "The amount of fruit varies from season to season because of weather, insects and temperature fluctuations. Producers want to know how much to expect and typically how they do it, is by looking at the trees before harvest. This could help them see crop yield ahead of time."

Such robots could also cut grass, trim trees, or spray for pests as they roam. Singh doesn't know whether the robots his team will develop will do several things at a time or specialize in certain tasks. But he's happy to show another way that Carnegie Mellon roboticists are developing applications for use outside of heavy industry and defense. "I think this is a fantastic way of contributing to society," he says.
For More Information: 

Jason Togyer | 412-268-8721 | jt3y@cs.cmu.edu