With labor at a premium, robotic fruit picker shows promise
With labor at a premium, robotic fruit picker shows promise
Almost four years ago, the California-based Citrus Research Board embarked on a project to determine the feasibility of developing a robot that could roam a citrus orchard and pick fruit distinguishing both size and quality in the process.
Ted Batkin, president of the board, said that the project reached a major milestone in mid-June that has advanced the concept from theoretical to reality. Now it is seemingly only a matter of time before a robotic harvester is picking oranges in the California's San Joaquin Valley as well as apples in Washington's Wenatchee Valley.
"It has become extremely realistic," Mr. Batkin said.
From day one, Mr. Batkin said that the board has progressed very conservatively and deliberately. Each step along the way, the idea has had to prove itself feasible, both practically and financially, before moving to the next step. He estimated that it is about a 20-step project and "we are on number seven or eight. But this last step was an important one. It was a critical control point. Steps eight to 18 should progress very rapidly as they are mostly mechanical engineering steps."
The latest milestone involved a practical application of using a robotic arm to take pictures of citrus fruit in a potted tree and then feeding those pictures into a computer to create a three-dimensional picking pattern for each tree. Mr. Batkin explained that the robotic harvester actually involves two separate machines. One machine will "scout" the grove mapping the trees and their fruit. The harvester will then be programmed to follow suit picking the designated fruit as determined by a computer program.
Mr. Batkin said that the scout machine can work in tandem with the harvester or it can do its job two weeks in advance. He said that once a tree has been mapped and a three-dimensional picking pattern established, the harvester would work independently.
Although it would seem that the tree would change as the fruit grows, Mr. Batkin said that is not really the case. And in any event, the computer program that has been developed can anticipate change such as growth and accommodate that change. It also anticipates that as fruit is picked and a branch's load is lightened, the branch will stick higher in the air, slightly changing its location on the tree.
The Citrus Research Board executive said that the program is designed so that once the tree is mapped, the robotic picker will look for patterns. It knows the approximate place for a 10-orange pattern, for example, and will find that group and pick the fruit according to the specifications entered into the program.
Mr. Batkin said that about $700,000 has been spent on the project so far, which includes close to $300,000 from the Washington apple industry. He said that about $500,000 more of industry money has been earmarked for the project, and then private donations of about $5 million will be needed to bring the theoretical idea to fruition, which will include building as many as 20 harvesters.
Mr. Batkin said that the final product has passed the cost-analysis test. "That is the first thing we did," he said. "Before we spent any money developing the computer programs, we spent a full year analyzing the costs. It has been analyzed on a cost-per-bin basis, and it does work."
He said that the cost analysis was done figuring that a pair of machines -- the scout and the harvester -- would cost about $350,000. "The cost of steel has gone up since then, and so it is probably closer to $400,000 now," Mr. Batkin said.
He expects that the first machines will be bought by packingsheds, picking contractors and other entrepreneurs in much the same way that private entrepreneurship fueled the development of mechanical harvesting in the wine grape industry.
"I know. I bought one of the first grape harvesters," he quipped.
The mid-June test was extremely important, as it was the first time that the project moved from theory to reality. Mr. Batkin explained that until now, the Citrus Research Board and its development partner, Vision Robotics in San Diego, have been developing the concept in virtual reality.
"That's the way you do things now," said Mr. Batkin. "Before we commit any money to building the machine, we have to prove that it works in virtual reality."
Until now, the development effort has revolved around creating the concept - the development of a picking pattern -- and the execution within the confines of the computer. Virtual reality showed that it could be done theoretically. Mr. Batkin said that the mid-June test proved that it could also be done with a real tree.
Now it is a question of money and time. He predicted that the funds will be raised and that the scout machine will be ready for commercial production in 18 to 24 months, with the robotic harvester following suit within the following 12 months. Thus, Mr. Batkin believes that there will be robotic harvesters in the groves for 2010 or 2011 crops.
While there have been some press reports that the robotic harvester is one answer to the immigration issue and the need for workers, Mr. Batkin downplayed that connection.
"Initially, we expect that ground crews will work together with the robotic harvester. The ground crews will harvest the bottom of the trees while the robotic arms will pick the top of the trees," he said.
Mr. Batkin said that possibly the best argument for such a machine is the worker-safety advantages. Currently, workers climb 20 feet in the air on ladders, pick fruit and deposit it in picking baskets. Unfortunately, accidents are not rare.
But both Mr. Batkin and Derek Morikawa, president of Vision Robotics, agree that the labor issue was at the core of the original idea.
Mr. Morikawa said that the company is working on a number of projects in agriculture, including the development of a robotic pruning device for grape vineyards. He said that agriculture is looking at accomplishing some tasks in the field mechanically in an effort to get control of its labor costs and to stabilize them.
Mr. Batkin said that the successful development of mechanical pruning and harvesting aids could help U.S. growers compete against foreign competition that has access to cheaper labor.
The Vision Robotics executive said that producers of many different agricultural products, such as apples, oranges, pears, cherries and cantaloupe, are exploring the use of robotic harvesters. Once the concept is perfected for citrus, he said that adapting it to a wide range of crops would be the logical next step.
In setting the timetable for seeing a robotic harvester in the citrus groves, Mr. Morikawa is a bit more conservative than Mr. Batkin, expecting that it will take at least 48 months -- which could still be in time for the 2011 harvest. But he said that a grape vineyard pruning robot, using much of the same technology, could be in use within 24 months.
And the "scout" robot, which can map the fruit in an orchard, could be helping the citrus industry accurately estimate its crop size as soon as the 2009 season.
Ted Batkin, president of the board, said that the project reached a major milestone in mid-June that has advanced the concept from theoretical to reality. Now it is seemingly only a matter of time before a robotic harvester is picking oranges in the California's San Joaquin Valley as well as apples in Washington's Wenatchee Valley.
"It has become extremely realistic," Mr. Batkin said.
From day one, Mr. Batkin said that the board has progressed very conservatively and deliberately. Each step along the way, the idea has had to prove itself feasible, both practically and financially, before moving to the next step. He estimated that it is about a 20-step project and "we are on number seven or eight. But this last step was an important one. It was a critical control point. Steps eight to 18 should progress very rapidly as they are mostly mechanical engineering steps."
The latest milestone involved a practical application of using a robotic arm to take pictures of citrus fruit in a potted tree and then feeding those pictures into a computer to create a three-dimensional picking pattern for each tree. Mr. Batkin explained that the robotic harvester actually involves two separate machines. One machine will "scout" the grove mapping the trees and their fruit. The harvester will then be programmed to follow suit picking the designated fruit as determined by a computer program.
Mr. Batkin said that the scout machine can work in tandem with the harvester or it can do its job two weeks in advance. He said that once a tree has been mapped and a three-dimensional picking pattern established, the harvester would work independently.
Although it would seem that the tree would change as the fruit grows, Mr. Batkin said that is not really the case. And in any event, the computer program that has been developed can anticipate change such as growth and accommodate that change. It also anticipates that as fruit is picked and a branch's load is lightened, the branch will stick higher in the air, slightly changing its location on the tree.
The Citrus Research Board executive said that the program is designed so that once the tree is mapped, the robotic picker will look for patterns. It knows the approximate place for a 10-orange pattern, for example, and will find that group and pick the fruit according to the specifications entered into the program.
Mr. Batkin said that about $700,000 has been spent on the project so far, which includes close to $300,000 from the Washington apple industry. He said that about $500,000 more of industry money has been earmarked for the project, and then private donations of about $5 million will be needed to bring the theoretical idea to fruition, which will include building as many as 20 harvesters.
Mr. Batkin said that the final product has passed the cost-analysis test. "That is the first thing we did," he said. "Before we spent any money developing the computer programs, we spent a full year analyzing the costs. It has been analyzed on a cost-per-bin basis, and it does work."
He said that the cost analysis was done figuring that a pair of machines -- the scout and the harvester -- would cost about $350,000. "The cost of steel has gone up since then, and so it is probably closer to $400,000 now," Mr. Batkin said.
He expects that the first machines will be bought by packingsheds, picking contractors and other entrepreneurs in much the same way that private entrepreneurship fueled the development of mechanical harvesting in the wine grape industry.
"I know. I bought one of the first grape harvesters," he quipped.
The mid-June test was extremely important, as it was the first time that the project moved from theory to reality. Mr. Batkin explained that until now, the Citrus Research Board and its development partner, Vision Robotics in San Diego, have been developing the concept in virtual reality.
"That's the way you do things now," said Mr. Batkin. "Before we commit any money to building the machine, we have to prove that it works in virtual reality."
Until now, the development effort has revolved around creating the concept - the development of a picking pattern -- and the execution within the confines of the computer. Virtual reality showed that it could be done theoretically. Mr. Batkin said that the mid-June test proved that it could also be done with a real tree.
Now it is a question of money and time. He predicted that the funds will be raised and that the scout machine will be ready for commercial production in 18 to 24 months, with the robotic harvester following suit within the following 12 months. Thus, Mr. Batkin believes that there will be robotic harvesters in the groves for 2010 or 2011 crops.
While there have been some press reports that the robotic harvester is one answer to the immigration issue and the need for workers, Mr. Batkin downplayed that connection.
"Initially, we expect that ground crews will work together with the robotic harvester. The ground crews will harvest the bottom of the trees while the robotic arms will pick the top of the trees," he said.
Mr. Batkin said that possibly the best argument for such a machine is the worker-safety advantages. Currently, workers climb 20 feet in the air on ladders, pick fruit and deposit it in picking baskets. Unfortunately, accidents are not rare.
But both Mr. Batkin and Derek Morikawa, president of Vision Robotics, agree that the labor issue was at the core of the original idea.
Mr. Morikawa said that the company is working on a number of projects in agriculture, including the development of a robotic pruning device for grape vineyards. He said that agriculture is looking at accomplishing some tasks in the field mechanically in an effort to get control of its labor costs and to stabilize them.
Mr. Batkin said that the successful development of mechanical pruning and harvesting aids could help U.S. growers compete against foreign competition that has access to cheaper labor.
The Vision Robotics executive said that producers of many different agricultural products, such as apples, oranges, pears, cherries and cantaloupe, are exploring the use of robotic harvesters. Once the concept is perfected for citrus, he said that adapting it to a wide range of crops would be the logical next step.
In setting the timetable for seeing a robotic harvester in the citrus groves, Mr. Morikawa is a bit more conservative than Mr. Batkin, expecting that it will take at least 48 months -- which could still be in time for the 2011 harvest. But he said that a grape vineyard pruning robot, using much of the same technology, could be in use within 24 months.
And the "scout" robot, which can map the fruit in an orchard, could be helping the citrus industry accurately estimate its crop size as soon as the 2009 season.
