BY the year 2000, a farmer should be able to buy his own agricultural robot for the price of a new pickup truck.
Researchers at Purdue University in West Lafayette, Ind., and the Volcani Institute in Bet Dagan, Israel, have devised a robot that can cultivate crops such as lettuce, cabbage, pumpkins, and melons. A single robot will be able to transplant, cultivate, and harvest, says Gaines Miles, a Purdue agricultural engineer and one of the inventors of the prototype.
``The robots will probably look like tractor-pulled trailers with grippers,'' Mr. Miles explains, ``but for the farmers with only a few acres of vegetables or fruits, they'll become as useful and versatile as a tractor.''
The advent of intelligent machines capable of selective harvesting has the potential of raising the quality of fresh produce, lowering production costs, and reducing the drudgery of manual labor.
The prototype has been developed to pick melons or any other head crop, because of the importance of these crops in the United States.
Despite advanced development in automation and electronic sensing, quality sorting of agricultural products is predominantly done by hand. Evaluating and sorting fruits is labor-intensive and not very reliable. For example, farmers currently determine the ripeness of a melon with their eyes, based on what they know from experience. Human failure, fatigue, and inadequate experience causes inconsistency in selection. A significant amount of harvested fruit is immature or overripe. This reduces the quality of marketed fruit and results in large losses to the grower.
To assemble the agricultural robot, Miles and several graduate students began with what looks like the skeleton of a large utility trailer. The eyes of the robot are cameras mounted on the trailer frame which scan the plants below.
A fan blows across the plants to move their leaves and expose hidden fruit. A computer then analyzes the camera images and looks for round, bright spots that might be melons.
To help differentiate a melon from other round, flat objects, a sensor projects a laser beam onto the ground.
When the laser strikes a spherical object at the same time the camera records a bright spot, the computer determines its ripeness through another sensing device. The sensor measures the level of aromatic gases coming from the melon. The amount of gases is directly related to the fruit's ripeness - the more gases, the riper the fruit.
Finally, the robot picks the ripe melons with its gripper arm. The gripper gently grabs a melon, lifts it, and cuts the vine.
``It could be outfitted to do even more,'' Miles says. ``While the grippers are transporting a fruit to the conveyor ... they could weigh it and add a bar code that tells the weight, variety, and harvest date.''
Funding for the project comes from the Binational Agricultural Research and Development fund, a US and Israeli fund. The prototype costs around $75,000, but Miles says manufacturers will be able to use lower-cost components in the future.
``Electronic components for an agricultural robot could cost less than $2,000 in the next decade,'' he says. ``Robots will be priced comparable to pickup trucks - well within the reach of farmers with small acreage.''