Robots can be trained to spot disease and estimate yield
From plows to seed drills to tractors, evolving technology has brought about radical changes to agriculture over the years. Now the sector is poised for another shift as robotic farmhands gear up to make agriculture greener and more efficient.
Three things now make mobile agricultural robots a real possibility in the near future, says Tony Stentz, an engineer at Carnegie Mellon University’s robotics institute.
Firstly, mobile robots have now proved able to cope with complex outdoor environments; secondly, the price of production has fallen; and, finally, society should now see robot laborers as a benefit not a curse.
Robots could address growing concerns in the developed world about a lack of labor availability in a sector reliant on intense bursts of tough, seasonal work. “Automation is becoming a necessity rather than an enhancement,” says John Billingsley at Australia’s National Centre for Engineering in Agriculture in Toowoomba, Queensland.
Swords to plowshares
Perhaps ironically, the fact that robots are now becoming capable of taking on the muddy challenges of food production is in large part down to the military.
The technology needed to make the leap from autonomous robots transporting things around factories to getting their wheels dirty in the field has been honed by events like the US DARPA grand challenge, a series of races for autonomous cars that had teams sending them across the desert or even through urban streets with real traffic.
Stentz helped an SUV called “Boss” win the Urban Challenge in 2007 and also worked on Crusher, a 6-tonne vehicle capable of driving unaided across extreme terrain.
“If you can deal with an off-road environment you have never seen before then you’re well equipped for agriculture,” says Stentz. “We have hit the elbow in the curve for this technology making it big outdoors.” He thinks that the next few years will see rapid changes in what robots can practically and affordably offer farmers.
But while having robots navigate their way through groves of trees may be similar to previously tackled robotic problems, getting them to read a crop like a seasoned rustic is another matter.
Stentz is experimenting with sending autonomous mobile robots along the rows of a Florida orange grove. The 3D laser ranging scanner used for navigation can capture detailed measures of every tree’s foliage and even count the oranges they bear, he says. His Carnegie Mellon colleague Sanjiv Singh is also gathering laser-ranging data in apple orchards in Pennsylvania (see image, top right).
Singh has also modified an orchard platform – a vehicle that drives along lines of trees carrying workers aloft to reach high fruit – to drive without human control. “Speciality crops like citrus, apples and other fruit trees have the most to gain from automation because they have not seen the same improvements in efficiency as other crops,” he told New Scientist.
Tree-reading machines could record data more often and more thoroughly than humans, providing early warnings of disease and more accurate yield predictions, says Stentz. This could help make the spraying of chemicals more targeted and efficient. “Instead of spraying at one constant rate, we can use [a robot-built] map to work out how to put down the minimal amount of chemical,” he says.
Robots that navigate using laser ranging can also work at night, when more insects are active and winds are less strong, Stentz adds, making chemicals go further. Engineers at Carnegie have already demonstrated that a robotic tractor can pilot itself around an orchard spraying water.
Such smart automation could take off if, like any new kind of consumer technology, it can offer enough benefits to be attractive to a large number of farmers, says Billingsley, and the costs should now be low enough. He is currently working on a computer vision system that monitors the condition of cattle on huge Australian cattle stations when they cue up to access a water hole.
The Carnegie projects also involve researchers from Florida, Cornell, Penn State, Washington State, Oregon State and Purdue universities, and a range of industrial partners.
Via New Scientist