Carbon Robotics’ AI Model Revolutionizes Weed Detection in Agriculture
What defines a plant as a weed, desirable or undesirable, is largely subjective and often down to the discerning eye of the farmer. However, this age-old practice is being revolutionized by advanced technology, specifically a new AI model developed by the innovative Seattle-based company, Carbon Robotics. This company is known for their LaserWeeder, a fleet of robots equipped with lasers designed to eradicate weeds. Now, their latest innovation, the Large Plant Model (LPM), is set to redefine how weeds are identified and removed, offering a more efficient and accurate solution for farmers worldwide.
Understand the Large Plant Model (LPM)
Announced on Monday, the Large Plant Model (LPM) is a cutting-edge AI model that instantaneously recognizes plant species. This allows farmers to target new weeds without the need to retrain the robots, making weed management more efficient. The LPM is trained on a vast array of data, including more than 150 million photos and data points collected from over 100 farms in 15 countries where the robots currently operate. This advanced model now powers Carbon AI, the artificial intelligence system that serves as the brains behind the company’s autonomous weed-killing robots.
From Time-Consuming Retraining to Instant Recognition
Before the introduction of LPM, each new type of weed encountered on a farm required the creation of new data tags to retrain the robots, a process that could take up to 24 hours each time, according to Paul Mikesell, founder and CEO of Carbon Robotics. With LPM, this time-consuming process is eliminated as the model can instantly learn and identify a new weed, even one it has never seen before.
“The farmer can live in real time and say, ‘Hey, this is a new weed. I want you to kill it,’ and that’s something that’s never been done before,” Mikesell explained in an interview with TechCrunch. “There is no need for re-labeling or recycling as the LPM understands, at a much deeper level, what it is looking at and what type of plant it is.”
Experience and Expertise in AI
Carbon Robotics, founded in 2018, started developing this model soon after it began shipping its first machines in 2022. Mikesell, with his years of experience in building these types of neural networks through his previous roles at Uber and his work on Meta’s Oculus virtual reality headsets, has been instrumental in the development of this groundbreaking technology.
Continual Refinement and Future Prospects
The new LPM will be implemented into the company’s existing systems via a software update. From there, farmers can instruct the machine on what to kill and what to protect by selecting photos collected by the machine via the robot’s user interface. With over $185 million raised in venture capital from backers like Nvidia NVentures, Bond, and Anthos Capital, Carbon Robotics is poised to continue refining the model as machines continue to feed new data into the LPM.
“We now have over 150 million labeled plants in our training set,” Mikesell said. “We now have enough data 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 structure looks like, without even having ever seen that particular plant before, because we have so much data coming into the neural network.”
The future of weed management in agriculture is promising with technological advancements like the Large Plant Model. As Carbon Robotics continues to innovate and refine its technology, farmers worldwide will increasingly benefit from more efficient and accurate weed identification and removal. For more information about Carbon Robotics and their AI model, click Here.
