A new generation of autonomous robots is assisting plant breeders in shaping future crops.
FARMER CITY, IL — Last fall, rows of 40-foot-tall maize shone in a study field along Highway 54. The University of Illinois at Urbana-Girish Champaign’s Chowdhary, a horticultural engineer, descended to set a little white robot on the border of a row labelled 103. All-terrain wheels and high-resolution cameras on either side of TerraSentia’s body gave it the appearance of an upgraded lawnmower.
By throwing out hundreds of laser beams to scan the surrounding area, TerraSentia works similarly to self-driving automobiles in “seeing” its environment. To get the robot in the right direction, a few taps on a tablet were all needed. The robot squeaked somewhat as it passed over ditches in the field.
Dr. Chowdhary remarked, “It will measure the size of every plant.”
This would accomplish that and more. To help agronomists in the future, the robot is meant to create the complete picture possible of a field, including the size and condition of the plants, as well as how many ears each grain will yield at the end of the regular season. For instance, TerraSentia’s “stand count” metric may list the number of grains or fruit-bearing plants, stem thickness, leaf number, and “stand count” metric all at once. In addition, Dr. Chowdhary and his colleagues at EarthSense, a spin-off firm he founded to produce additional robots, are attempting to add even more features, or phenotypes, to the listing.
Hand measurements of these phenotypes have long been the norm among plant breeders, who used them to choose plants with the most exemplary traits before crossing them. While the discovery of D.N.A. sequencing has assisted, it still requires a human to determine if the genes separated from the prior generation have genuinely led to an improvement in their offspring’s features.
A rise in the number of bots
Automation of the phenotyping procedure using robots, according to Dr. Chowdhary, would increase the reliability of measurements. Using technology like TerraSentia, farms can achieve yield optimization levels previously unattainable by people alone.
From the earliest seed drills to today’s agricultural machinery, farming has relied heavily on mechanization. Detecting weeds and calculating herbicide application rates using deep learning and robotics are just two examples of the many uses for sensors on farm equipment.
Smaller, more agile robots have been popping up all over the place recently. Oz, a three-foot-long, 300-pound robot developed by the French firm Nao, made its debut in 2014 with the release of ten prototypes. While eating weeds, it collects phenotypes of edible crops. In Switzerland, EcoRobotix has developed a solar-powered robot that can identify crops and weeds in a matter of seconds. Another soil and plant analysis robot, BoniRob, has been tested by Bosch, a home appliance manufacturer.
“All of a sudden, people are beginning to realize that data collecting and analysis techniques established during the 1990s technological boom may be used in agriculture,” said Carnegie Mellon University systems scientist George A. Kantor, who is utilizing his research and developing techniques for calculating agricultural yields.
The TerraSentia is the tiniest farmbot currently available. The 30-pound robot, which is 12.5 inches broad and nearly the same height, can squeeze between rows of crops. Data from earlier in the agricultural supply chain is also collected as a primary emphasis. Varieties chosen for commercialization are selected in these research plots.
TerraSentia’s data is transforming breeding from a reactive to a predictive process. In the same way that physicians use genetic testing to evaluate the possibility of a patient acquiring breast cancer or Type 2 diabetes, scientists may use the robot’s sophisticated machine-learning skills to collect the effect of hundreds or even thousands of elements on a plant’s future features.
Mike Gore, a plant scientist at Cornell University, says that “phenotyping robots can select the best-yielding plants before they ever release pollen.” As a result, the time required to generate a new cultivar—the result of selective breeding—could be reduced from eight to four years.
Creating a Market
Agriculture’s global need is increasing. According to the United Nations, the human population will reach 9.8 billion in 2050 and 11.2 billion in 2100. With less land, fewer resources, and in the face of climate change, farmers will need to increase their technical intelligence to feed the globe.
The agricultural titans are interested in learning more about the technology. Currently, the TerraSentia is being tested by Corteva, a spin-off of Dow Chemical and DuPont that was formed in 2016.
Neil Hausmann, Corteva’s director of research and development, remarked, “There is absolutely a niche for this sort of robot.” The standardized, objective data it gives helps us make many judgements. This information assists us in determining which products are best, which should progress, and what qualities are most suited for producers in different regions of the country,” he says.
For Dr. Chowdhary and his colleagues, the robot subsidies will come through collaborations with large agribusinesses and academic institutions.
Chowdhary said farmers do not need any specific knowledge to run the TerraSentia. Nearly all of the robot’s functions have been transferred to it. A farmer in a developing nation with just five acres of land may use a single unit to survey his crops just as readily as a farmer with thousands of acres of land. For example, the TerraSentia has previously been evaluated in various fields such as maize and soybean, sorghum, cotton and tomato, wheat, strawberry trees, citrus crops, almond fields, and apple orchards, and vineyards.
On the other hand, some experts doubt that such robots would ever be aimed at small farms or be an economical alternative. According to Kyle Murphy, a policy and agricultural development specialist at M.I.T.’s Abdul Latif Jameel Poverty Action Lab, “there are many impediments to the adoption of new technology for the sort of agriculture that smallholders prefer to participate in.” According to the professor, robots like TerraSentia may be more likely to benefit smallholder farmers indirectly by encouraging the creation of better or more suited crops.
Improvement is a long and continual journey
Until TerraSentia has perfected its crop-breeding talents, it cannot help farmers across the country. To correct the rover’s trajectory after it has tripped over branches and debris on the ground, the user must go behind the rover and push it back into place. Dr. Chowdhary hopes that by next year, TerraSentia will be able to be trained so that even more users will not need to be in the field.
For the time being, the TerraSentia travels at a slow rate of less than one mile per hour. It enables its cameras to detect even the tiniest changes in pixels to measure the plant’s leaf area index and detect indicators of illness. EarthSense’s Dr. Chowdhary and his colleagues are banking on improvements in camera technology to speed up the robot in the future.
As part of this project, the TerraSentia will have a power socket built for when it is through working for the day. At this location, the vehicle’s battery may be replaced with a fully loaded one, and its axles and sensors can be cleaned. However, for the time being, farmers toss the robot into the back of a truck, drive it home, and upload its data to the cloud for further analysis.
EarthSense’s headquarters in Urbana, Illinois, is filled with prototypes of robotic technology that never materialized. First-generation TerraSentia models had no suspension mechanism, so when researchers deployed them in a muddy area, they would bounce and interrupt the camera feeds. They changed the polymers in the robot’s motors to avoid melting and added metal insulation.
Those early, damaged frames are now piled on a shelf, like in a museum: a reminder about the need for progress and the enthusiasm generated by the robot.
Some of those who tried the original designs returned home to us as well after getting robots that effectively broke down on all of them and the way, Dr. Chowdhary added.