Developing Artificial Intelligence for Agriculture
Artificial intelligence has the potential to unlock new opportunities for agriculture. But in order for this technology to be effective, the model needs to be trained to accurately identify what it’s seeing. So for companies like TerraClear, an agtech company whose first product is a robotic rock picking arm, they picked a needed use case to invest in developing their artificial intelligence model.
Thompson… “At a very basic level, we're saying, okay, what is the widest use case where people need to have in our case, you know, a map of, of rocks but it could be weeds or it could be really anything else.”
That’s TerraClear president Trevor Thompson.
Thompson… “And so let's get that one right first. And then let's sort of see, can we have a, transfer learning or other technical approaches that allow us to then add different field conditions that maybe have a little bit of a smaller market.”
Thompson says for artificial intelligence to really take off in ag, there needs to be a shared database that companies can access to train their models more quickly.
Thompson… “I think this is the challenge for solutions that are using AI in many cases is it is a tremendous amount of data, and we've got narrow windows. You know, so if you're looking at a specific weed that only really happens in a very kind of tight window. So data sharing or some sort of marketplace where we can kind of exchange, but we have to figure that out or else a lot of these solutions are going to take years because you're just really limited.”
Developing technologies that actually work in the field is no easy task.