Driverless vehicle progresses
The goal is to increase efficiency and reduce costs in fruit production.
The autonomous prime mover utility vehicle is guided by laser sensors that sense the distance from the vehicle to the trees.
A team of scientists has demonstrated that it’s possible for a vehicle to drive itself back and forth between rows of trees in an orchard. The next steps are to make it easier for growers to control the equipment and to convince them that this is not science fiction, but something they will find useful in their own orchards to increase their efficiency and reduce costs.
The scientists, led by Dr. Sanjiv Singh at Carnegie Mellon University’s Robotics Institute in Pennsylvania, began two years ago to develop what the university refers to as an “autonomous prime mover” as part of a larger, multistate research effort on Comprehensive Automation for Specialty Crops.
Carnegie Mellon has developed autonomous equipment in the past, but primarily for the U.S. Department of Defense and the National Aeronautics and Space Administration. Singh said the goal for the first year of the project was to study the feasibility of using autonomous equipment in specialty agriculture, recognizing cost considerations.
The scientists started out by converting a Toro two-seater utility vehicle into an autonomous electric system. They recognized that they couldn’t depend on a Global Positioning System to guide the vehicle between tree rows because highly accurate GPS systems are expensive, and reception under dense canopies is poor. To achieve the same accuracy as a high-end GPS system and still make the vehicle economically feasible, they used laser sensors that can sense the distance of the vehicle to the trees. Not only can the vehicle drive between rows of trees and turn at the end of the rows, but in wide rows it can be directed to drive closer to one side or the other.
In the first year, the team logged a total of 120 kilometers (75 miles) of autonomous operation, exceeding the goal of 100 kilometers.
Having proven the feasibility in the first year, the scientists focused last year on making the system robust enough to work in orchards or vineyards that might have uneven terrain, dense or light canopies, and extreme temperatures. It was tested on rolling hills at the Fruit Research and Extension Center in Biglerville, Pennsylvania, as well as on slopes and flat ground in Washington State. On flat ground, the laser beams can sense up to 30 meters (100 feet) ahead, but on hilly terrain, they sense the ground in front.
The goal for the second year was to have the vehicle drive at least 100 kilometers, but only runs of at least 10 kilometers at a time counted. That meant it had to operate continuously for hours at a time, navigating numerous tree rows. Singh said this was accomplished, although there were real-life challenges to address, such as dust clouds during which the lasers picked up reflections from the dust and confused the software.
The term “autonomous prime mover” is a concept, rather than a specific vehicle, Singh said. During the early development phase, they’ve been using the Toro utility vehicle, but last year they used the same navigation concept with an orchard platform from Italy that they converted from a manual motorized vehicle to driverless.
The team just began the third year of the project with the goal of making the autonomous prime mover easier for growers and farm employees to operate. It will have a graphical user interface (somewhat like an iPhone or iPad) so that its operation is more intuitive. The idea is that the user would simply command the vehicle to start at one end of the block and execute a certain task—spray herbicide, for example—and off it would go, and then return when it was done.
This year, the team hopes to have the autonomous prime mover log 100 kilometers of autonomous travel when operated by people other than robotics engineers. The system will be simplified and rebuilt without the research instruments that are currently used to monitor its performance. “We’ll know how well it does by watching it,” Singh said. “It will look simpler, and be simpler and more robust and easier to use.”
Modes of operation
Singh said an autonomous vehicle should be able to serve a variety of purposes. His team has identified four modes of operation:
Transport mode: This is when someone drives it manually from the shed to the orchard block where it’s needed, for example.
Mule mode: People load the vehicle and it drives itself to some other location—perhaps the end of the row—where it is unloaded and then it returns. It could also be used as a “bin dog,” where it follows a person on foot wherever they go.
Pace mode: The vehicle drives up and down rows spraying or mowing without people on board.
Scaffold mode: The vehicle has someone on board working on the crop as it drives slowly down the row.
This year, Karen Lewis, Washington State University Extension educator in Washington’s Columbia Basin, and Dr. Tara Baugher, with Pennsylvania State Cooperative Extension in Adams County, will both test driverless utility vehicles.
Lewis said she is having a two-person platform with electric scissor lift installed on the back of the vehicle and will test it for all types of orchard work except harvest. Growers are interested in orchard equipment that is modular and multipurpose, she said, and the two-seater utility vehicle is useful because it can carry two people and a lot of gear. “I think we’re going to see more use of these vehicles,” she said.
Singh said the final year of the project is likely to focus on the economics of the autonomous prime mover, and growers still need to be convinced of the utility of driverless equipment.
Growers who have seen the driverless utility vehicle demonstrated at field days have described it in such terms as “beyond amazing,” but they probably don’t fully appreciate yet what it could do for them on their farms, he said.
The Comprehensive Automation for Specialty Crops project is funded through the U.S. Department of Agriculture’s Specialty Crop Research Initiative. •