Robotics scientists at Carnegie Mellon University in Pennsylvania set out to develop a device to automatically measure the caliper of nursery trees in the field, but found that a simple tree counter would be more useful for the tree fruit industry.

Dr. Sanjiv Singh and colleagues began to develop the tree caliper calculator with shade tree nurseries in mind. The difference in value between large- and small-caliper trees can be significant. The device would be attached to a vehicle and would assess tree size as it moved down the nursery rows.

The project is part of a major research project on Comprehensive Automation for Specialty Crops.

When the researchers discussed the concept with tree fruit nurseries, however, they found that being able to automatically count the number of trees would be more important than being able to measure the caliper in the field.

Singh said developing a counter was much easier than figuring out how to measure the caliper. He and his colleagues first tried an ultralow-cost approach using an off-the-shelf infrared diode that cost about $25. However, the infrared was susceptible to interference.

They then purchased a commercial, off-the-shelf sensor in a sealed unit that cost between $300 and $400 and developed the electronics and software to process the data so the results could be viewed on a laptop computer.


Paul Tvergyak, marketing director with Cameron Nursery in Eltopia, Washington, said tree fruit nurseries sell trees by caliper, but they could benefit most from a fast method of accurately measuring caliper and bar-coding the trees after harvest, rather than in the field, to provide an accurate inventory of what’s in storage.

In terms of monitoring in the field, a tree counting device would be the most useful initially because nurseries count trees several times between planting and harvesting for inventory control, he said.

“Counting trees is one of the most difficult things I have ever done. They’re planted six inches apart in rows a quarter of a mile long, so you’re counting along and if you get distracted by something you’re thinking, ‘Where was I?’ You’d think it would be a pretty simple deal to count them, but it’s not.”

The first count is done when the rootstocks are planted in March or April. The second is when the planting crew reports what they actually planted and in which row. The trees are counted again when they’re budded in August. In September, the trees are counted for the fourth time to see which buds survived. The fifth time is the following spring to find how they survived the winter. The last live tree count is done in September and the seventh when the trees are dug in November. In addition, the nursery might do another half-dozen partial counts to track specific variety or rootstock combinations. “All the time, we’re checking,” Tvergyak said.

The numbers change from count to count, particularly if a block is hit by a fireblight infection and replacement trees have to be planted. Once a fireblight outbreak has been stabilized, the trees are counted once more.

The reason for keeping an up-to-date inventory is to know how many trees the nursery has to sell and avoid overselling, he said. “You don’t really know what you have until you have them in hand.”

An automated tree counter would save a tremendous amount of work, he said. “If you could come up with some automatic thing, you could take less qualified people and have some confidence in the ­numbers.”

A tree counter that could also map the location of trees by variety and rootstock, perhaps using a Global Positioning System, would be particularly useful, Tvergyak added.