How a Machine Learning Robot is Revolutionizing Solar Farm Maintenance-OnSight Technology Interview

How a Machine Learning Robot is Revolutionizing Solar Farm Maintenance-OnSight Technology Interview

Show Video

Billy Fetzner: On solar panels, we have all these different devices that are monitoring different things about the solar panel itself, determining how healthy that solar panel really is. But sometimes all of that data is just data and it is really difficult to really diagnose what's actually happening on the solar panel. And that's where OnSight Technologies come in, because sometimes it's just better to see it. They have a little robot that goes around the entire solar farm and is able to take thermal images and pictures of these solar panels to see, you know, what's exactly going on. And it's like having. Your own remote technician that is constantly there on site.

Pretty incredible, right? Well, there is so much more that this robot does. So make sure you check out this entire interview with the CEO of OnSight Technologies, Derek Chase, because he goes in-depth to tell you why every solar farm needs this robot. Derek, do you mind just telling us a little bit about your company OnSight Technologies and what you guys do? Derek Chase: Yeah. OnSight Technology it's a robotic inspection tool used for utility solar plants, and you're able to basically operate it from anywhere in the world, but it cruise around some of these really large solar farms and gather an immense amount of data about the solar panels and how they're operating and the site in general, other electrical equipment on site as well. It has a thermal image camera and a 4k, a hundred X optical zoom camera on it as well.

And we're continuing to add more sensors to gather more data. And the idea is basically these farms are hundreds of acres, big and there's really no viable option now to gather the really detailed information you need to understand if something detrimental or catastrophic is gonna happen onsite. Billy Fetzner: Definitely.

Yeah, that's really cool. And so you guys have a little robot that goes around in each of the rows and goes up and down, right? Derek Chase: Yeah. Basically it's a Roomba for really large solar farms that, and it has a camera on it and other sensors and they gather data and allow us to better operate these really large solar farms, which can range into like billions of dollars of asset. Billy Fetzner: Definitely.

On an average solar farm, how long does it take your robot to go and do its rounds? Does it do multiple times a day or is it, very, just depending on the size? Derek Chase: Yeah. We're riding a variety of technologies in terms. The battery capacity, how fast you can charge is obviously a big deal. And anybody who's dabbled in the electrical vehicle market understands some challenges there. So it has an inductive charger that it docks up to, and we're thinking it's probably four hours on four hours off cycles.

So you should be able to get 12 hours of like basically the robot working and gathering data per day. And we think that adds up to around between 30 and 50 megawatts a month of really detailed inspection. Billy Fetzner: Nice. Very cool. And so how does that compare to what they would have to do otherwise? So instead of, sending a team out there versus, using this. Derek Chase: Number one, we just finished up a demo in Arizona where it was it was like 110 degrees outside.

And so to have somebody go walk around 110 degree weather and take pictures and document things was really difficult. Let alone electronics won't even stay on. Often times your phone will turn off and you can't gather the data or take the pictures.

So it's really avoiding putting people in the field unnecessarily and gathering a bunch of data before you have to send them out there. So when they do go out there, it's really cost efficient and it doesn't put them in dangerous situations. Billy Fetzner: Definitely. What temperature does the robot do good at being able to operate? Does it have a maximum temperature? Or is it able to go with the environment? Derek Chase: It's been on site and tested in areas like I think it was in Modesto for a while up in the Sacramento area for over the past year, it's been in the field. It done very well.

It's our test robot. And it's handled the temperatures really well. Like the cameras outdoor rated. It's just it's like a really sophisticated security camera. That's outdoor rated, so they're made to work in extreme temperatures. And the battery is shaded by our power shell.

And so it seems to operate fine so far. We're really gonna do our inspections in the morning and in the evening when the modules are tilted and then easier to inspect. So we can also inspect at different times of the day. So it doesn't get beat up by the heat too much, which typical work days are somebody's gonna go work.

They're gonna work during the daytime hours. And that's typically when it's hot. Billy Fetzner: Definitely and so What are the, the biggest sellers of the types of data that you're able to gather from the robots? Derek Chase: Yeah.

So it has a thermal image camera and that allows us to look at the solar panels and the solar panels aren't working as good as they should be, or there's an issue with them. They'll actually show it through. The resistance builds up and it gets really hot.

They call 'em hotspots in the jewelry industry. So the camera can see those hotspots and we can start to see if there's an issue with one panel. All the panels typically came from the same manufacturer at the same time. So we can start to see a group of trend over time. Is there anything we can do about it? Do we need to replace the panel? And we gather a lot of this information to going, what actions do we take next? Or maybe all the panels are good. And we actually feel really good about the site and actually increases the value of it and takes away, some of the uncertainty.

I think that data's valuable either whether it's good or bad, and , we're able to look at the solar panels, the junction boxes on the back of the solar panels, where all the wires are connected. That's like a spot that there's been some failure. Also the connectors where the modules actually the solar panels are all connected together. Those are a failure point as well. And we're able to see all of those from the back of the panel, with the robot.

Billy Fetzner: Definitely. And how does this compare to, because you have an actual robot there, being able to check out these things compared to like on the unit centers, sensors that are just there to. Make sure everything is functioning correctly, but, compare that to your robot was able to do even more, right? Derek Chase: Yeah. We're able to give a very real time visual perspective of what's going on at the site.

And I think that's pretty valuable. Everything else is just like sensors, where they gather data, but it's a lot of numbers. It's tough to sort through and sometimes. I was the analogy that I always used is if you had a house or an apartment, you were renting it out to somebody and you just looked at the utility bills and gathered the rent checks. And you did that over a period of five years.

You think, Hey, everything's fine at this house here based on the data I'm receiving, but then you actually go look at it and it could be completely destroyed. Your asset's now worth half the value. They gotta make all these repairs. So I think there's something pretty valuable about The ability to just see what's going on out there and see the condition of things, whether that's rust or hotspots or some of these like physical issues that can happen on the site. Billy Fetzner: Definitely.

And how does the robot, when it's doing its rounds, does it give a notification to the developer to let them know Hey, this looks like an issue. Like you should look at this, or how does that go? Derek Chase: Yeah. So right now the operator can do two modes, basically. There's like a mode where it will just automatically cruise around and take pictures of things.

And then all those pictures and images gathered go through our machine learning program. And this is really of the value behind the robot is The visual machine learning technology we have developed this allows all these pictures to be analyzed. And then basically it says, Hey, there's this a good picture, or has good things in it, or as expected.

And, or this is a bad picture and this has hot spots or defects that we've identified. And as we continue to sort those, the machine learning program gets smarter and smarter. It can start to identify. The issues, it will geotag those issues. So it'll let you know where they're at the time of day that the picture was taken, et cetera, all the other metadata that comes with it.

And that can go into a report and send to the operator on a daily basis from a general inspection. The other way is the operator can actually drive the robot around and we use an Xbox controller for this. You got like basically zero latency. It's really impressive. It feels like you're like literally at the site you can zoom in on things. So maybe a high level engineer, if there's some issue they want to take a look at, they can just put themselves on site virtually and do this inspection.

And with that, we have an easy click report button to where they do find something. They click a report and there's all the metadata and stores all that for them to look at a later. Wow. That's, Billy Fetzner: that's really incredible. I'm just making it so much easier. to sort through that, all that data that comes in.

And then also to be able to like, say like you went and did a repair and then you're coming back later and saying okay, instead of having to go out there and constantly check is it still working fine? You can just, use some robot and go specifically, go to that site and say, this is where we fixed it. Oh, it looks fine or whatever. Derek Chase: Yeah. One of their use case we actually found during our last demo in Arizona was Somebody's working on a piece of like really technical equipment, like some of these really big inverters its a MegaWatt, and you can actually have the technical expert in the phone, in the robot, you can have some speaker on a microphone, so you can talk to it.

And so the person on the other end can zoom in and really feel like they're there and give the technician on site, a lot of sound advice on how to complete the repair and make sure everything's working appropriately. So it really brings the office staff into the field. To help out onto complex issues. Billy Fetzner: That's awesome that, that's really great, especially when if it's really hot outside and your phone dies and you can't use it and all that kind of stuff. So that's really good too.

Wow. That's awesome. And so what is your mission with OnSight Technologies? Derek Chase: Yeah, really the holy grill is 100% automation and a variety of different sites to where you could just drop it off. And it goes into Roomba mode basically, which is pretty tough when you consider all the different like site conditions that can be out. There's one big, like self driving, operating component, how you control the camera, et cetera.

And then the other one is a really great machine learning backend to that, that can analyze all the predict features and produce the reports. So that's really what we're striving to. We believe that once we get there there's a variety of other use cases beyond just solar plants that we can apply this to cuz we're in a lot of interesting places at that time.

Whether. Electric vehicle with machine learning with autonomous driving. So we'll continue to look for different use cases there, but we've got long ways to go to make sure we tap out the utility solar market first. Billy Fetzner: So how would you apply this to the EV market? Derek Chase: No, just in terms of our robot is an electric vehicle, right? So battery power. So once we can inductively charge it, control it, drive it around. What other use cases can that, can there be, is there like agricultural uses seems to be like a pretty close relationship with some of the solar farms that we're doing.

So whether that's inspecting for, with the, are the vegetables ripe or other weeds or different things, that's a pretty big market as well. It'll take us a while to really crack the code on all of the different types of utility. And different use spaces our test clients we've been working with are throwing all kinds of different use cases who is right now.

So we seem to have a lot of different targets to go after and a lot of value to create. But beyond that, we think the problems we're solving today can be applied to some, a lot of other industries in the future. Billy Fetzner: Yeah. I got you now.

That's very cool. You'd have to manipulate your algorithm and things, but you still got, the base background of the machine learning type of idea what you're analyzing and those kind of things. You'd have to relearn different aspects of, different pictures to apply to, different categories of whichever new market you're going into.

But that's definitely very cool. Derek Chase: Absolutely. Yeah. It's all exciting technology, whether it's tele operations, which is really interesting.

Just controlling something from somewhere else. I think is just it's so much fun. I just, I, my son drives the robot around and can do infections he's seven years old. He can do it from our house. So it's really.

Getting people in the market to understand the technology that's really available and how to use it. That's really been the challenge because not everybody's used to those capabilities. So you've got a few hurdles there Hey, not only can there be a robot on site, but you can actually drive and manipulate it in real time. And then not only that the data can be gathered and gone through a machine learning program. Is there like a few steps ahead from where the industry is today. A lot of it's just like educating the market and making sure that they understand how it all works.

And then how it applies and then how that's really valuable to them. So there's a few steps to get in it. Billy Fetzner: Yeah. But there's a lot of potential there. That is really cool. And what is like the return on investment for a solar developer who is, going to apply this OnSight Technology with the robot and everything.

What is their return on investment and how quickly can they expect that? Derek Chase: Yeah so number one, if you buy the robot with a new utility solar site it qualifies for the federal tax credit. You can get up to 20% off on the asset plus accelerated depreciation. It's an asset and it goes on your balance sheet. So it shouldn't really be negative cash flow or hard to achieve an ROI.

As soon as you start saving on some of the truck roll or your ability to prevent more new hires there's a few different PNLs we're going after. One of the PNLs is just the overall cost of manage a solar side. You have a 50 megawatt solar site. And because we put a robot on it, it's now cheaper to operate that site and less risky that site's now worth more money.

So we're improving the value of the assets to reducing the operating costs and then the companies themselves that operate those assets operations of maintenance companies, if you're able to reduce their operating costs they're having less personnel or less truck rolls or better data and limiting power loss. Then you're also improving the value of those operations of maintenance companies. So we think everybody's winning in the supply chain. It makes sense. You put it on your balance sheet.

It's an asset. The goal's really I think the ROI would be less than 12 months. And we have a variety of different ways that the customers can purchase it, whether that's buy it or lease it or rent it for a particular cast, et cetera. It should be a win and a no brainer all around. Billy Fetzner: Definitely sounds like it.

And what do you foresee as being like the biggest challenge for you, helping developers understand that, this is a valuable and how do you look to overcome those kind of challenges about, knowledge and understanding? Derek Chase: I'm not sure if I'm gonna be able to convince anybody. And there's a lot of noise in general in the news and the market today that except to get people's attention to maybe look forward, but we have found if you're really forward looking companies and that I've been impressed with in terms of their willingness to invest in innovation and get on the front end of it. I think it'll be painstakingly obvious maybe too late for some folks that adopt it because at a certain point we won't be able to just produce these, for everybody all at once. So I think the early movers get a lot of advantages of building out the technology specifically for their needs. So we're trying to show the value in. Some, it will just be shared.

They won't be able to get the people that they need. The sites won't be able to be inspected. I think some of it'll just be the sheer market demands and evolution of the problem that will drive the adoption. So that may take a little bit longer for some folks to actually feel the pain before they have to make the move.

And I think some of those early innovators that jump on board early and craft the product for themselves will be able to see some huge savings. Billy Fetzner: Definitely for sure. And what are your goals for OnSight Technologies for the rest of 2022? Derek Chase: Right now we're finishing up the autonomous driving, finishing up the machine learning. These are constantly in development and super deep wells.

We've got a really smart team. And our PhD, John Shepard he's our CTO and heads up our team. What's a few engineers, right out of the Davis area. And they've had some up with robotics in the past. So we're just trusting them on doing their thing and continue to support them. They've made tremendous progress in a short time.

We've been going, so definitely finish up on those technologies. We've got a variety of different really large partners that we're looking to select one to really go to market with and deploy a lot more units to start gathering data faster. Just keep things moving forward and find some really strong partners early on. We don't want to get caught in operating a really overwhelming operating company and an R & D company, we want to finish up our R & D while working through some strategic partners early on. Billy Fetzner: Definitely. And who would be the perfect strategic partner for you? Derek Chase: We've got a couple we're talking to right now.

So we're gonna see who comes to the table first, but basically it's somebody who owns or manages solar sites. To where we can deploy the technology to, but what's really more important is like an engaged group that gives us feedback to help us mold the product and then share with us some of their financials. So we can have really good use cases on how we can improve their profitability. And we think ultimately, that's, what's gonna move the needle for.

Billy Fetzner: Yeah, for sure. If they can see, okay, I do this and this is the outcome I received, they can definitely make that correlation very easily. And what about more of a general public on, how can they support your efforts and what would you suggest that, just anybody who's interested in this kind of stuff what would you suggest they might be able to do. Derek Chase: Yeah, I think we're building like a, it's just a really fun project and we're welcoming. I've been really open with our goals and our technology and how we're using it.

So everybody's really welcome to follow along the journey. We post a lot of cool stuff on our LinkedIn site. And we're gonna continue to build out the YouTube side of things, but I think machine learning is just so interesting. Just the black magic behind it, and it's gonna, how it's gonna evolve over these next five years.

I think it's gonna really make a huge impact. I've had so much fun learning about it over the past year from my team and the capabilities that's been really impressive. I think wherever you can find a company that's utilizing that. Technology's really interesting. And we're really open.

We've got some partners with micro technology who help us out. We've got a really impressive board we've put together some really smart team, really impressive board. And we're bringing a lot of people along for the ride, so we'll see how it goes. Billy Fetzner: Definitely and what is one tip that you would give to another green business? ecopreneur out there? Who's looking to grow their business? What would you suggest for them to do? Derek Chase: You ask a good question. You've gotta make sure you're solving a problem. That brings in value to the situation.

There is an ROI for the product. There's a big enough market. Somebody else hasn't done it before we were able to join an accelerator program that was really helpful for us that helped me build a network and get things off the ground, be a part of the community early on. So yeah there's some really interesting startups in the space coming around and you're starting to see all these new technologies being used, whether. Energy storage is a huge place right now.

And energy management, a lot of software, energy management companies. A lot of ESG kind of stuff out there. So yeah, great time to get in there. A lot of different new technologies to play with, to solve problems. So I think there's not a small list of problems to solve.

Billy Fetzner: No, definitely not. And what is one way, if somebody is interested in reaching out to partner with you or interested in just learning more how can they reach out? Derek Chase: Yeah, I think, a lot of this stuff we're doing is through LinkedIn right now. I think we're B2B company. That's an awesome B2B marketplace. I've really enjoyed participating on there and watching some of the other companies I'm the fan of. Definitely reach out, follow us and send me a message and we'd love to connect and you, we can work together.

Billy Fetzner: Yeah, definitely. And I know you mentioned you're into learning about machine learning right now. What is somewhere else that you're, into learning and learning about currently? Derek Chase: Yeah, I think we'll obviously like LIDAR technology, autonomous driving. New GDS in general are really interesting on where they're gonna go.

Cause the capabilities, when you start combining all they things together are really interesting. Tele operations. If you've played Oculus, have you done Oculus? Billy Fetzner: I have, yes. It's amazing.

Derek Chase: yeah. It's you're playing ping pong in real time with somebody across the world. It feels like that it's that same experience when you're tele operating a piece of equipment, you're gonna start putting those things together.

I thoughts a link online today where somebody was in an Oculus, controlling a robot somewhere else. I think that's gonna be a big deal here is people controlling things from somewhere else. And of virtually being able to be there. So all those technologies are super interesting to me and we're incorporating all of them, so hopefully we can ride that way into something impressive. Billy Fetzner: Yeah definitely. And make an impact on the world.

Thank you so much for coming today on onto this show. It's been really awesome having you, I'm really excited about seeing OnSight Technologies grow you guys are headed in the right direction and really gonna make an impact on solar developers. So I think it's really awesome what you're doing.

Derek Chase: Yeah. Thanks so much for having me. You got a great show here, love the impact you're making on educating people about different companies and technologies and all the effort you put into this. So yeah, I really appreciate your. Billy Fetzner: Thank you, Derek.

Thank you so much. And if you enjoy this interview with Derek talking all about their machine learning technology and the awesome little robot that goes around on these solar farm sites, then I invite you to check out this interview with Sunified. They are using their own machine learning technology and putting that into a decentralized system so that you can possibly get your own little piece of the sun through Sunified.

2022-09-04 16:45

Show Video

Other news