A clever combo of CV, lasers, and autonomous nav bring weed removal into the 21st century.

The news

Seattle-based Carbon Robotics has announced its latest weed-zapping robot, the Autonomous Weeder.

The agricultural tool uses computer vision and autonomous navigation to identify weeds as it moves. Then, it uses precise, high-powered lasers to zap weeds as it moves. Autonomous Weeder is also designed to work at scale—it can run 24/7, cover 15-20 acres a day, and destroy up to 10,000 weeds per hour.

Why does it matter?

Weeding is a dirty and dull job to perform manually. That’s why farmers started using chemicals to reduce weeds and grow their production. But chemicals can be quite harmful in agricultural applications. They damage the crops themselves, reduce yield, and trickle into the food chain.

Autonomous Weeder promises the best of both worlds. It offers the environmental benefits of manual weeding without the need for a huge labor pool. And it offers the scalability of chemicals without the deleterious environmental effects.

In fact, Carbon Robotics says the tool is certified organic. And that means it could offer an easier and much cheaper path to organic certification for a variety of farmers.

How it works

The Autonomous Weeder is a culmination of significant technological development in CV, autonomous navigation, and sensor production.

As the robot moves, it feeds data from 12 hi-res cameras into Nvidia processing hardware. This on-board computer uses computer vision models to identify crops and weeds. Next it uses 150W lasers, arranged in 8 modules, to zap the weeds.

For autonomous nav, the robot uses front and rear cameras, as well as CV models trained to recognize furrows. Like other autonomous agriculture robots, it moves to the end of a row, turns around in a safe zone, and then heads down the next row. Should any obstacles get in the way, the robot uses lidar sensors (presumably 2D) to recognize and respond appropriately.

For more information, see Carbon Robotics’ website. Or just check out the video below.