An Interview with Gabriel Torres: AI, Agriculture and Drones

Gabriel Torres

As part of our AI in Your Community series, I sat down with Gabriel Torres, an unmanned systems expert and CEO and co-founder of MicaSense, which brings unmanned systems and sensors to the agricultural industry. We discussed his work with MicaSense and how they’re working to help the agricultural industry take better advantage of tools and technologies.

Tara Chklovski: What problem are you working on?

Gabriel Torres: In general, optimization. Ever since I was in school I’ve been interested in efficiency. I see inefficiencies in the world, and I see opportunities for things to be easier, better, more transparent, more universal. I always like to see how I can impact somebody’s life for the better. I’ve done that throughout my career.It’s been really rewarding, to really make an impact and to touch lives in ways that I would normally not be able to.

TC: Tell me more about MicaSense.

GT: With MicaSense we started to make it easier for the agricultural industry to catch up with and take advantage of technologies that haven’t really been available or accessible, and provide a benefit, in terms of quality, quantity, and yield.

The Parrot Sequoia is a powerful multispectral sensor. Via Micasense

TC: How do your products help with that?

GT: Our product is a multispectral sensor, which is a small camera that can be carried on board aircraft – usually a drone but also manned aircraft. It captures different spectrums of light, in a very precise and calibrated and quantitative manner, so it becomes more of a scientific sensor than a camera. Information from the sensors is processed and analyzed to provide information about what’s going on in the field. For instance, they can tell us about stress in crops even when that stress might not be visible.

TC: Cool! The AI Family Challenge asks families to find a problem in their community and develop some sort of technology (potentially with some sensors) to collect data; so, very similar to what you are doing, but on a simpler level. What advice would you give children when they are trying to find a good problem to solve using technology?

GT: That’s a great question. I try to find a problem not just for the sake of solving a problem, but something  that is causing pain, costing money, costing time, affecting the environment…sometimes those things aren’t immediately obvious. I would tell children and families to identify something that speaks to them, a problem that they themselves have a personal connection with and that is significant to them. And sometimes those problems are not solvable, at least not now, and sometimes those problems are solvable.

TC: And what advice would you give about designing a good product?

GT: There should be always a perspective of, who’s going to use this?  Many times, when we look at solving problems, especially on the engineering side, we seem to get a little bit carried away. We forget that there are real people at the other end of devices or applications that are using the product, and we don’t perceive how they’re going to use it. And we end up with a product or a technology that is incredibly innovative and new, and has all sorts of benefits, but can’t be used effectively or cannot be scaled. A good rule of design is to start at both ends. You start by defining the problem. You start by defining what the possible solutions to that problem are. And then you imagine that you have the solution for the problem and you go all the way to the end, and figure out how the person that is going to benefit from this device or product or software is going to use it. Are they going to be able to use it at scale? Do they need to use it at scale? Do they have the capability to use it? I’ve run across that multiple times in my career and even at MicaSense, where we imagine the solution to a problem and then we realize it’s a great solution, but a very small number of people will actually be able to use it. So, then you have to go back to the drawing board.

TC: Stepping back a little bit, what inspires you?

GT: I’m inspired by doing things that change lives and perspectives. I’m inspired by solving problems that haven’t been solved before or solving them in a different way, in a way that that is transcendental, that matters. I guess everyone has that desire in the end. I know that’s something that’s special for me.I’ve had the privilege of being able to do that across my career, and I feel that that’s something that I want to continue, that inspires me to be able to touch people’s lives, to affect them in a positive way, to give them tools, to share with them the things that I know how to do that can help their lives be better.

TC: What do you find difficult, and how do you overcome difficulties?

GT:  I find it difficult sometimes to grasp all the unknowns. That’s one of the hardest things in engineering and design: you define a problem, or a solution or technology, but there’s this overhanging cloud of unknowns, things that you do not know you need to know, questions that you haven’t asked because you haven’t thought of asking them. And as you start actually getting your hands dirty and solving the problem you realize that those things are popping up, and sometimes those things are critical, they’re catastrophic. Meaning that they’re not able to be solved. And you never thought that they needed to be solved. So, the uncertainty of what’s not known is something that I’ve always found difficult because it’s hard to move forward without it.How do I cross that difficulty? Part of it is being willing to accept that I do not know everything and that there are things that we do not know and that we have to proceed with some level of risks. Flexibility is a key part of it, flexibility to change and say, “okay, this didn’t work, but let’s try it this way”. And I think a big part of it is surrounding myself both personally and professionally with incredibly smart people that complement the things that I know how to do and that are way smarter than me on many, many parts of the problem and taking their advice, taking their knowledge, and trying to form together a complete solution.

TC: What field of AI excites you the most?

GT: Well, as a result of my work I’ve become very interested in the vision side of AI. I’ll give you an example. My daughter Eva and I are working right now on a little project for Perler Beads. One day we were playing with them and they got all mixed up. We started trying to sort them out, and I said, we could do this practice program, with a chip, a computer. And she was somewhat intrigued by that. So we went out and bought a little camera with a chip in it that has OpenCV (Open source Computer Vision library) and Open Machine Vision libraries and we’ve been working on a little bead sorter.Of course, as you can imagine, eventually my six‑year old daughter got kind of bored with it, but the project remains in my head. To me, those kinds of problems exist all over! At her school, they have a bucket full of these Perler beads and they’re all mixed up, and wouldn’t it be cool if we could help them sort? You could spend hours and hours and hours sorting through these things or you could use technology to help you do that and spend your time doing more fun things.

TC: What is the best way for children to learn about AI?

GT:  Well, you know what I’m going to say to that. And I imagine you would have exactly the same answer, which is to go out and do it. The best way for anybody to learn is to go out and do it, to try it.

TC: We are super excited for what the different families come up with after we share basic tools and kits and some knowledge of how to use it, but invite them to go build what they think we need. And so, we are hopeful that we will a get a lot of interesting variable things around health, or crop health or crime, air quality, things like that. We feel like over the next year things are even going to get more and more accessible and cheaper and more powerful. The technology’s really moving fast.

GT:  You’re absolutely right, Tara. One of the things that will probably help spur this is more and more companies are becoming much more open in the sense of becoming more universal and having their technologies used which didn’t happen before.We’re doing that at MicaSense. We’re seeing benefits from it. We’re seeing that by sharing the technology that at one point was guarded with lock and key, it encourages innovation that benefits everybody and benefits companies financially.

TC: Interesting. The technology’s open, but the data is so valuable because that’s what needs to be generated by human beings, collected, analyzed, understood, and acted upon.

GT:  The AI is not just capturing the data well, but the AI is making sense of that data. That’s the other lesson learned: in some cases there’s too much data. And not enough people, not enough capabilities to make sense of it so it just gets stored and it never gets used.And that’s not good either.

TC: Right. And I think that’s a hard part, because the AI agent has to be trained on what an important feature or a pattern or a signal is.

GT:  Our brains do that. They take all these that signals and what we would call data and rather than drawing a complete three‑dimensional model as you’re walking around the room ,they tell you “stop, you’re about to bump into something.” We need AI that does the same. We need the AI not just to capture the data, but we need AI to make sense of the data and translate it into something that people care about and can use. And this goes back to the question of how do you do good design and make sure that people can use what you build. How do you make data that people can actually use?This is where AI and robotics and technology marry really well with human intelligence and experience. I mean again, taking agriculture as the thing that I’m most familiar with, when we talk to our customers and people that are looking at the technology we see it as a tool, a tool in the belt of the people who really understand what’s going on with a particular crop or field.It requires interpretation. And that interpretation is not something that a machine can ever learn in my opinion. That interpretation requires knowledge of, say, what happened a decade before, that might not necessarily be recorded or requires knowledge that of what has and hasn’t worked on a specific farm. And that’s not necessarily the information that is really available. So, we see this information has to be actionable. Actionable information shouldn’t be blind.

TC: Right.

GT: We should treat AI and computing in general as ways to provide complimentary information that we use around intelligence and experience and insight and intuition to make a combined decision. And we see that in agriculture. We see our products and our data products being used in conjunction with soil data, with weather data and with knowledge of the field, and with an agronomist’s intuition to make a recommendation in the end. It’s not a one‑stop shop.

TC: Humans alongside machines.

GT: And I think this is part of what you’re trying to achieve is to make that point clear and in our case every time that we give an educational talk, there are a lot of misperceptions about, well this is going to tell when me what I need to do– where do I put fertilizer, or exactly what kind of disease I have… it’s yes, and no. It will tell you information that helps you make that decision, but the technology cannot and should not be used on its own for those kinds of decisions, because we’re talking about living things in this case. It’s not so robotic.

TC: Right. All right, Gabriel. Thank you for sharing all these hard-earned insights, and for opening up the world of agriculture to our families. We’re very excited about seeing what kinds of interesting products people come up with. Thank you for opening that door.