WHY THIS MATTERS IN BRIEF
A model that learns any plant on sight lets farmers ditch herbicides and retraining for instant, targeted weeding.
Matthew Griffin is the World’s #1 Futurist Keynote Speaker and Global Advisor for the G7 and Fortune 500, specialising in exponential disruption across 100 countries. Book a Keynote or Advisory Session — Join 1M+ followers on YouTube and explore his 15-book Codex of the Future series.
What is and isn’t a weed that needs to be eliminated in the field is determined by the eyes of the farmer — and now, increasingly, by a new Artificial Intelligence (AI) model from Carbon Robotics.
Seattle-based , which builds the LaserWeeder — a robot fleet that uses lasers and not increasingly expensive herbicides to kill weeds — announced a new AI model, the Large Plant Model (LPM), on Monday. This model recognises plant species instantly and allows farmers to target new weeds without needing to retrain the robots.
The LPM is trained on more than 150 million photos and data points collected by the company’s machines across the more than 100 farms in 15 countries where the robots currently operate. The model now powers Carbon AI, the AI system that serves as the brains inside the company’s autonomous weed-killing robots.
The Future of Food Keynote | Hoogstraten, Belgium | Matthew Griffin | Futurist Keynote Speaker, by Futurist Keynote Speaker Matthew Griffin
Paul Mikesell, the founder and CEO of Carbon Robotics, told reporters that prior to LPM, every time a new type of weed would show up on a farm — or even the same type of weed in different soil or with a slightly different appearance — the company would have to create new data labels to retrain its machines to recognise the plant.
This process took about 24 hours each time, Mikesell said. Now, LPM can learn a new weed instantly, even if it’s never seen it before.
“The farmer can live in real time and say, ‘Hey, this is a new weed. I want you to kill this,’ and that was something that had never been done before,” Mikesell said. “There’s no new labelling or retraining because the Large Plant Model understands, at a much deeper level, what it’s looking at and the type of plant.”
Mikesell said that the company, which was founded in 2018, started developing this model shortly after it began shipping its first machines in 2022. Mikesell has years of experience building these types of neural networks from previous roles at Uber and working on Meta’s Oculus virtual reality headsets.
This new model will reach 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 photos that the machine has collected in the robot’s user interface.
Carbon Robotics has raised more than $185 million in venture capital from backers including Nvidia NVentures, Bond, and Anthos Capital, among others. Now, the company will look to continue to fine-tune the model as the machines continue to feed the LPM new data.
“We have over 150 million labelled plants now in our training set,” Mikesell said. “We have enough data now that we should be able to look at any picture and decide what kind of plant that is, what species it is, what it’s related to, what its structure is like, without having ever even seen that particular plant before, because we have so much data going into the neural net.”
Why does instant plant recognition matter for farming?
Until now, every new weed — or even a familiar weed in different soil — meant a roughly 24-hour cycle of relabelling and retraining before the machines could act. A model that generalises to plants it has never seen removes that bottleneck, so farmers can point at something in the field and have the robots act immediately, cutting both herbicide use and downtime.
