Several manufacturers are watching this innovative "see weed, spray weed" technology closely, preparing to bid for the rights to this smart sprayer that could redefine proper chemical application and save big bucks for farmers.

The double-decker booms reaching out one side of a self-propelled sprayer draw odd looks from farmers, especially if the onlookers get close enough to see video cameras shooting the ground from an 11-ft.- high boom.

Surely this isn't a lawyer's warped advice to custom applicators to negate claims that they haven't been adequately covering fields. No, actually it's the latest prototype of the University of Illinois' smart sprayer. And it's bringing closer to reality the ability to accurately spray only the weeds that are present.

Eliminate broadcast. You know your fields and how erratic weed distribution is over a whole field. Common sense says that broadcasting a contact herbicide would be efficient only if you were able to treat single weeds and patches.

According to Dr. Lei Tian, ag engineer and developer of this machine-vision smart sprayer, only 20 to 30% of a normally applied post-contact herbicide actually kills weeds. "We've designed the smart sprayer to increase this efficiency as close to 100% as possible, with the goal that it will save farmers thousands of input dollars while reducing environmental and crop stress load.

"In fact," Tian says, "a 50 percent savings should be easy to get since, conservatively, only 50 percent of a field has weeds."

After testing their first sprayer and computer software system in 1997, the engineers proved that the concept worked. The "real-time" system sensed weeds, then triggered individual nozzles to spray these weeds with a water and dye mix, with 83% accuracy.

How it works. This complex system begins with live video footage to identify weeds. Currently, one camera watches a 12-ft.-wide area and controls six nozzles (20-in. spacing). The camera feeds images to the computer, which distinguishes weeds from crop, residue and bare soil. It can recognize weeds compared with crop plants by analyzing leaf shape, size and texture.

Upon receipt of the images, the software signals the appropriate nozzle(s) to spray exactly when the weeds pass under the spray boom. The current prototype also can vary the chemical rate being applied determined by weed size and density all while maintaining a constant boom pressure and traveling 6 to 7 mph. Tian says that the technology is probably capable of handling speeds up to 10 mph.

Ag engineers Tian, John Reed and Brian Steward, have vastly improved the 1998 prototype, added to a Tyler Patriot XL sprayer, by using the latest video camera, sensor, nozzle and computer/software technology.

Their goal is to increase accuracy and bring the design closer to production standards so it can be mass produced at a reasonable price.

"We've made many improvements to increase performance," Tian says. "Our computer and software designs are now Windows-based to give immediate visual results. Older near-infrared cameras have been replaced with color, multispectral, high-resolution cameras (placed higher off the ground) to detect weed size and density. We've tied our software and the sprayer controller that we designed into a GPS/GIS system to generate weed maps for future use. We're working on adapting the software to environmental conditions, such as clouds and lighting conditions, which can affect weed identification."

Obviously, the crop canopy cannot get too large or this system won't see enough ground. "But the crop is normally smaller during the proper application window for post control of weeds," Tian says.

Although this rig has yet to apply a herbicide, all tests have been validated using a special dye in water to determine spray coverage on weeds.

Ready for market? Typical of most researchers, Tian says he'd like a few more years to refine the system. But with several manufacturers seriously looking at this technology, Tian hinted that a version of this smart sprayer could arrive in fields sooner than that.

But what about cost? Tian expects that, when mass produced, the machine can be reasonably priced. Because the nozzles are designed to be attached to any sprayer, the main expense will be the cameras and computer.

For more information, contact Dr. Lei Tian, Dept. of Ag Engineering, University of Illinois, 360-L Ag Engineering Sciences Bldg., 1304 W. Pennsylvania Ave., Urbana, IL 61801, 217/333-7534.