In Autonomous Russia, Tractor Drives You
Two Russian companies have teamed up to make an advanced autonomous driving system for agricultural vehicles. Rusagro, one of Russia’s largest agricultural companies, joined with Moscow startup Cognitive Pilot on a project they’ve named Cognitive Agro Pilot.
Cognitive Agro Pilot purports to be a level 3 autonomous system, which means a truly self-piloting system that requires virtually no human intervention. Most new consumer cars are level 1, so they have an assortment of “assistance,” but still require the driver to steer and otherwise interact at all times. At level 2, like Tesla’s Autopilot, vehicles can limit the need to steer or even to brake with the flow of traffic.
Agriculture is a perfect testing ground for fully autonomous “eyes off” level 3 vehicles, because large swaths of farmland need to be planted, treated with fertilizers, and so forth in very regular patterns. Farms around the world have faced a labor shortage for years, and that’s been rapidly accelerated by the global COVID-19 (coronavirus) pandemic.
But where more attention has sometimes been paid to skilled manual labor like picking, Cognitive Agro Pilot seeks to supplement a limited workforce for advanced farm machinery. By taking the element of driving out of the equation, workers can concentrate “more on managing and controlling other harvesting parameters,” says Olga Uskova, an AI researcher and Cognitive Pilot’s CEO.
Uskova told Robot Report the new system is unique for its reduced reliance on sensors and GPS, both of which lower operating cost for farmers. Instead of an array of sensors and cameras and other bells and whistles, the system uses input from just one camera to feed a custom neural network that helps plan ahead and adjust for obstacles like workers on the ground.
Instead of synthesizing several inputs that each handle separate ideas, this system parses one input for everything. From Cognitive Pilot’s website:
“The basic technology of the system is the analysis of images coming from on-board video cameras via computer vision algorithms. The system is able to ‘understand’ the types and positions of objects along the way, build trajectories and give control commands for the necessary maneuvers to be made. [A] detailed obstacle map is built up around the robot.”
Autonomous farming is a huge field of research and development, and this simpler, less costly system offers an interesting solution that’s super easy to implement and use. Partnering with Rusagro means the systems will roll out on more than 240 combines around Russia. “It’s noteworthy that the industrial introduction of the system fell in a year when a record harvest of wheat is expected in Russia,” Rusagro’s agriculture CEO, Roman Shkoller, told Robot Report.
Finally, Cognitive Pilot says allowing workers to concentrate on harvest tasks instead of driving will also improve outcomes. “In general, the use of the system can reduce the cost of grain by 3 to 5 percent and reduce its losses during harvesting by two times,” Uskova says. Once COVID-19 travel restrictions are lifted, she says the company will begin selling the system around the world.