Carbon Robotics developed an AI model that detects and identifies crops

What is and what is not a weed that needs to be removed from the field is determined by the eyes of the farmer – and now, even more, with a new model of AI from Carbon Robotics.

From Seattle Carbon robotswhich makes LaserWeeder – a robotic vessel that uses lasers to kill weeds – announced a new AI model, the Large Plant Model (LPM), on Monday. The model recognizes plant species instantly and allows farmers to target new weeds without the need to reprogram robots.

LPM is trained on more than 150 million images and data collected by the company’s machines on more than 100 farms in the 15 countries where the robots currently operate. The model now powers Carbon AI, an AI system that acts as the brain inside autonomous weeding robots.

Paul Mikesell, the founder and CEO of Carbon Robotics, told TechCrunch that before LPM, every time a new type of weed appeared on a farm – or the same type of weed in a different soil or a slightly different shape – the company had to create a new set of data to reprogram its machine to identify the plant.

That took about 24 hours each time, Mikesell said. Now, LPM can learn a new weed instantly, even if it has never seen it before.

Mikesell says: “The grower can be in real time and say, ‘Hey, this is a new weed.'” There’s no new label or retraining because the Large Plant Model understands, at a very deep level, what it’s looking at and what kind of plant it is.

Mikesell said that the company, which was founded in 2018, began to develop this model after starting to ship its first machine in 2022. Mikesell has many years of building these types of neural networks from previous positions at Uber and working on Meta’s Oculus virtual reality headsets.

The new version will arrive on the company’s existing systems through a software update. From there, the farmers can tell the machine what to kill and what to protect by choosing the images that the machine collected on how the robot works.

Carbon Robotics has raised more than $185 million in equity capital from sponsors including Nvidia NVentures, Bond, and Anthos Capital, among others. Now, the company is looking to continue to improve the machine as the machine continues to feed new LPM products.

“We have over 150 million plants listed here,” Mikesell said. “We have enough data now that we can look at any image and decide what type of plant it is, what species it is, what it’s related to, what its shape is, without ever having seen that plant before, because we have a lot of information going into the neural net.”