Working to make New Mexico’s chile industry high-tech and healthy
Across southern New Mexico and into Texas and Arizona, a major effort is under way to modernize the harvesting and production of a product near and dear to many lovers of Southwestern cuisine — chile.
Mechanization is coming gradually to an industry that has been synonymous with handpicking and hand-cleaning for many decades. Survival is now at stake.
Mechanization of this valuable crop ($300 million in New Mexico alone) is a critical step toward success, says Roy Pennock, a Cooperative Extension research specialist at New Mexico State University who has spent his life in the industry. As labor costs and availability fluctuate and availability of red chile powder from Peru, Africa, India, and China increases, the industry has come together to fight back. The New Mexico Chile Task Force combines growers, processors, crop consultants, university extension experts, and others “. . . to get everyone working on the same page,” says Pennock.
Add Sandia Labs to that mix, as Chris Wilson, Maritza Muguira, David Novick, Jon Salton, and Jesse Schwebach, all of Intelligent Systems Controls Dept. 15234, are working hard to play a contributing role. Now moving into the fourth year of a project with the task force, Chris and his team took their high-tech contribution to the effort on the road for last year’s harvest.
Working with Pennock, New Mexico State University Extension Engineer Ed Eaton, and others, Chris and the Sandia team have developed an imaging system that can measure the effectiveness of mechanical harvesters, cleaners, and sorters for chile producers. “We looked at the mechanical devices under development and decided we could help most with a measuring system,” says Chris. The system measures chile on a conveyor belt and quantifies the percentages of chile and “field trash,” which generally consists of sticks, leaves, and other natural debris.
The system was used in connection with a two-stage mechanical chile cleaner developed by Eaton during last year’s harvest. But it may also be of use to processors in the future, Pennock believes. “Mechanically harvested chile isn’t always perfect. You get pods, but you get leaves, branches, and maybe a few other things,” Pennock says. Without the “vision device” developed by Chris and this team, laborious before-and-after sampling is needed to gauge success.
“I think the system has tremendous potential for processing plants. You could have it at the beginning and at different stages of the system and it would tell you quantity and quality of the chile during processing,” says Pennock, who operated the system last year as part of the evaluation process.
The original idea of how Sandia might help the task force has evolved over the past three years, Chris says. He started out with the goal of doing a survey to provide some factual information on which methods of mechanical cleaning work and which do not. Originally, he thought Sandia’s robotics talents might be brought to bear on developing machines, but others, including researchers at NMSU, were ahead of the curve in this area, so Chris found another niche. He continues to consult with Eaton on design issues but has focused on measurement.
Cleaning chile fresh from the field is complicated by the fact that the peppers change throughout harvesting season. Early in the season the plants are green and fresh and there’s little field trash. Later as the plants turn red, mature, and endure frost, mechanical harvesters tend to pull up large amounts of brittle branches and leaves with the pods.
“Chris has been a good sounding board for me,” says Eaton. “He’s someone I can talk to who listens and is very helpful.” Eaton plans to take a new version of his cleaner, mounted on a conventional chile harvester, into the field this year. “We need investment to help us go forward to the commercial stage,” he says.
Examining different imaging technologies, Chris and his team developed a system that analyzes the chile and debris on a conveyor belt based on color differences. A digital camera connected to a portable computer takes still images of cleaned product on the conveyor belt. Software then analyzes the image, segmenting it according to color into product, trash, or background. Then the system counts pixels and provides feedback to the operator on percentages of product and waste. The operator can then adjust the cleaner and recheck the output plots to see the effect.
The project has thrown problems at Chris that he hadn’t seen before. “There are a lot more variables out in the field than there are in a controlled laboratory space,” he says. Given the variety of difficulties, Chris believes the task force, led by NMSU’s Rich Phillips, is doing a good job. “They’ve reduced the scope of the problem significantly. Part of what we do involves educating customers as well as trying to listen to them.”
Right now, most measurements are made by “eyeball,” says Chris. “There is no standard for estimating the amount of product. An objective metric system is needed.”
To achieve segmentation or determine what part of the image is actually the conveyor belt, debris, or chile, the system operator must “train” the software. The operator can develop appropriate masks to screen the images, based on hue and saturation values plotted as histograms. “This makes it easy to move from one conveyor belt to another with different color belts, or to measure differences in chile color based on the variety being harvested,” says Chris. “We should be able to work with our customers to make changes as necessary.”
“This work builds on and adds to what we are doing at Sandia. Here, I work on a lot of 3-D imaging, and this is a switch to 2-D. We are stretching ourselves in some different directions, but I think it will make us stronger.”