It is up to the farmer’s eye to determine which weeds need to be removed from the field and which do not. And now it’s up to Carbon Robotics’ new AI model.
Carbon Robotics, the Seattle-based company behind the LaserWeeder robots that use lasers to kill weeds, on Monday announced a new AI model called the Large Plant Model (LPM). The model instantly recognizes plant species, allowing farmers to target new weeds without retraining the robot.
LPM is trained on more than 150 million photos and data points collected by the company’s machines on more than 100 farms in 15 countries where the robots are currently operating. This model is currently powered by Carbon AI, an AI system that acts as the brain within the company’s autonomous weeding robots.
Paul Mikesell, founder and CEO of Carbon Robotics, told TechCrunch that before LPM, every time a new type of weed showed up on the farm, or the same type of weed in different soil or with a slightly different appearance, the company had to create a new data label to retrain the machine to recognize that plant.
Mikesell said the process took about 24 hours each time. Now, LPM can instantly learn even new weeds you’ve never seen before.
“Farmers can say in real time, ‘Hey, this is a new weed. I want you to get rid of this,’ which has never been done before,” Mikesell said. “Large-scale plant models understand the target and plant type at a deeper level, so no new labeling or retraining is required.”
Mikesell said the company, which was founded in 2018, began developing this model shortly after starting shipping its first machines in 2022. Mikesell has years of experience building these kinds of neural networks from his previous job at Uber and the development of Meta’s Oculus virtual reality headset.
tech crunch event
boston, massachusetts
|
June 23, 2026
This new model will be applied to the company’s existing systems through a software update. From there, farmers can tell the machine what to kill and what to protect by selecting the photos it collects on the robot’s user interface.
Carbon Robotics has raised over $185 million in venture capital from backers including Nvidia Nventures, Bond, and Anthos Capital. Going forward, the company plans to continue fine-tuning the model as the machine continues to feed new data to the LPM.
“We currently have over 150 million labeled plants in our training set,” Mikesell said. “We now have enough data, so much data fed into neural nets, that even if you’ve never seen a particular plant, you should be able to look at a picture and determine what kind of plant it is, what species it is, what it’s related to, what its structure is like.”
