AI is changing how and what we can recycle / Apurba Pradhan / Everest Labs

https://www.everestlabs.ai/

apurba@everestlabs.ai 

What is the future of recycling? 
How is AI making everything faster and more efficient with recycling? 
How can AI robotics and people work together to make the world a more sustainable place? 

https://podcasts.apple.com/us/podcast/packaging-today-show/id1656906367

Join Us Live Daily on LinkedIn Or YouTube or listen at your leisure on Apple or Spotify 

Packaging Today Podcast 

https://open.spotify.com/show/6dksVwqEFVDWdggd27fyFF?si=e924995740f94e19

https://www.linkedin.com/in/cory-connors/

I'm here to help you make your packaging more sustainable! Reach out today and I'll get back to you asap.

This podcast is an independent production and the podcast production is an original work of the author. All rights of ownership and reproduction are retained—copyright 2022.

Transcript

Cory Connors: 0:15

Welcome to Sustainable Packaging with Cory Connors. Today's guest is Mr. Apurba Pradhan from Everest Labs. He is the head of product. How are

Apurba Pradhan: 0:24

you, sir? Great. Thanks for having me on, Corey. It's good to chat with you and look forward to the discussion.

Cory Connors: 0:29

Oh man. for the people listening per Peru has this beautiful background behind him with, light shining through and it's like, like a halo. And it's such a nice change from the rain of recent. So, this is very exciting and, it's great to see you in this light. But let's talk about your background. what got you to where you are?

Apurba Pradhan: 0:49

Great. yeah. So, like you said, I head up products for a company called Everest Labs. We are a, quote unquote startup in Silicon Valley. my role is to look after sort of the product direction, and make sure that the company's sort of headed the right way, where our customers are headed and the market is headed. So we work in recycling, and sorting. And so it's important for us to stay, connected with folks who are in sort of in, not just directly involved in recycling itself, but also sort of the adjacent, industries like packaging and, package goods manufacturing, supply chain for circular economy. All of those things are important cuz they drive recycling companies. So we want to sort of stay ahead of those types of trends as well. So my job is really to sort of listen, understand and bring feedback back into the company to, to figure out what we should be working on.

Cory Connors: 1:43

That's excellent. it's not an easy task. this whole concept of recycling and sorting in, in the MIRFs, the material recycling facilities, can you tell us exactly what, Everest Labs does and their, what your products do? Yeah.

Apurba Pradhan: 1:59

So in, in short we, have an AI and robotics solution that MRFs, use to operate more intelligently their facilities, as well as, automate processes, right? So say, operate more intelligently. MRFs are pretty much a big factory. If you can think of it like a big factory without all the bells and whistles of a modern factory, right? They've got materials coming in, they're, they've got process steps where materials are being sorted and then they exit, as bales or go to landfill. And each of these, each of the steps in the process there, some kind of an industrial sorting system that's fairly. I guess cause industrial in nature and not as accurate as it needs to be. so We, install, camera systems, 3D camera systems. We use AI to understand what's happening at each step of those processes to help them opt, optimize those processes. at the end of those processes, typically there's a person, sorting out. Right. Things that those industrial extra sorting systems weren't able to, get. Because there's a, there's you need to sell bales with certain level of purity. and those jobs, are challenging. I think during covid it was actually a crisis. People weren't able to actually fill those jobs. Today it's not a crisis as much, but it's still a problem. Right? It's an aging workforce. It's hard jobs. and, it's hard to fill. People have to pay more and more money to people who are doing those jobs. So we help automate that by, by installing industrial robotics in those systems. So that's essentially what, Everest Labs does. But as in doing that, we capture a lot of data, and a lot of data, and we've got initiatives in the company to figure out, okay, how can we use this data to make not just the MIRFs better But just. recycling industry and the circular economy better as well. So, hopefully we get to talk about

Cory Connors: 3:51

that a little bit. Yes, I'd definitely like to cover that topic as well. I want to talk about the whole supply chain of waste and why is automation and AI and whatever's, labs, and you are doing so critical to the future of a successful and sustainable packaging world. let's talk. Why sorting materials is so difficult today and what Everest Labs has done to make it at least somewhat easier for a material recycling facility.

Apurba Pradhan: 4:26

Yeah, it's a good question. I mean, I think, why is it difficult? You kind of have to take a step back and say like, what's the state of recycling today? Right? Yeah. I think the state of recycling today is where it is because of sort of. Many things, but I think the whole idea behind single stream recycling, which most places in the US does, is there because it's what had worked, for consumers, households, through everything. Throw everything in a single bin, don't worry about it outta sight, outta mind, kind of think somebody else will take care of it. and usually taking care of it has fallen on the shoulders of, you know what, in the US here we call MRFs material recovery facilities, right? So they. They collect these items and they sort 'em and put 'em back into the supply chain. That's the, that's sort of the thesis. Yeah. I think recycling, if if you just say, Hey, recycling is like two big broad buckets, like one is sorting and collecting, and another is actually taking that material and reprocessing it to be virgin material. I think that latter part has had a lot of technological advancements, a lot of investments companies with new chemistry, new physics, new industrial processes, and there's some great technology out. To recycle, use material on the collection and sorting side not so much, right? the investments in technology, modernization of infrastructure, that area just has not gotten that much focus and MEFs. It's a very challenging business, right? it's a public partner, a public private, sort of a partnership. they're behold to contracts from municipalities, margins thin. they have to meet certain thresholds for sort of, how, what the, Amount of recycled content comes out of those facilities, but also have to meet a margin, right? they can only afford to staff and buy so much modern technology and staff. So much people. So they're really struggling with sort of the economics of collecting, sorting, and putting materials back into the supply chain. And that's probably the reason why. it's challenging. Not, technology exists, processes exists, but you can't just go and rip and replace an existing, merf facility. it's a very expensive proposition. Yeah. so I think that's kind of if I had to build a sorting facility from scratch, I could probably build a pretty good one. but there's already 400, 500 MEFs out there, you. Large to mid, medium size to large. If you count the small ones, there's probably like a thousand of 'em out there. But you know, there's probably 400, 500 MEFs out there and we gotta work within that existing infrastructure. Right, right. And that's. I think the philosophy that our founder jd, who started Aris Labs sort of, that was his philosophy as he went into this whole figuring out what is the technology needed for MEFs not to sort of say, Hey, let's come in here and kind of. Create a solution that is a big capital expense and it changes the entire system, but let's work within the current infrastructure. That's kind of, it's kind of, I think that's my long answer of why I think it's difficult to sort materials in merf

Cory Connors: 7:37

and very well said, our consumers like you and I want. Recycling to be easy and we want to have the ability to toss it in the blue bin, and and not worry about it and hope and pray that it gets into the right, merf and it actually gets recycled. it's impressive that companies like yours are making that dream come true. but what I'm seeing is even more sorting coming in the future. By, consumers. is, does that seem to be part of the potential solution to this?

Apurba Pradhan: 8:14

do you mean sorting directly in at the homes and sort of, locations? So, so definitely I think moving upstream helps the process. We don't do much, directly with consumers. We are looking at, we are working on solutions that are sort of still creators of, recycling, but not homes. So, businesses, large venues. Yeah. Stadiums, concert ev, and event centers. Multi-family properties, where actually the recycling rate is terrible.

Cory Connors: 8:42

what types of material is your, equipment really good at?

Apurba Pradhan: 8:46

Yeah. So, there's the AI part of it, then identification part and the sorting part. so on the AI side, AI had gotten really good at this point, we have something like 30 classes of objects, material classes that we detect like material classes, maybe something like P E T or H D P E or LDPE. And then under that, they're So there. Colored H D P E or natural H D P E or there might be PT bottles or Thermoform. there might be Tetra Pack and all the different classes of Tetra Pack, and it's all in all, there's about a hundred or so, classes of. Packaging that we can detect with like over 90% accuracy. Incredible, oftentimes over 95% accuracy. the AI is really good. on the robotics side, it's really good at containers, bottles, containers, those are sort of we, In our solutions, we actually guarantee success. so one of the things with robotics in sorting facilities and recycling is, there's a, there's a hurdle that we have to kind of get past, around trusting robotics Yeah. and new technology for a lot of these facilities and. you we're asking these operators take a leap of faith that a robot in AI that they know nothing about is going to do a job that, of a person that they've have 20 years managing, right? Or of a, a workforce that they have a lot of experience. managing training, kind of hiring things like that. So, so, we've had to sort of face that head on. and the way, one of, one of the ways that we do that is through performance guarantees. So basically we, guarantee that we can pick a certain number of objects, over, over a certain number of objects and we guarantee a success percentage of 80% in the picking, which is the highest in the industry. That's impress. Wow. If AI detects something and the robot will at least eight out of 10 times picket, and, typical that number is about 90, 93%, but we sort of put a floor at 80 saying, we'll guarantee that. So, so I think, yeah, the robotics are getting pretty good. I know obviously there's limitations. I mean, robotics can't go into these the the robotics, the current implementations of robotics. Can't. To picking bowling balls and cribs and things. All of these things that come in to sort of presort someday. Someday, yeah, someday. on the, what they call QC after, assorting equipment, the robotics have gotten pretty good. Exciting

Cory Connors: 11:15

to hear because. I know that the MFS have been really needing this for a long time, so well done to you and your team. I know when we talked before about this interview, your team mentioned, consumer product group brands wanting data from you, as to. What percentages or what information can you supply to them about their packaging and how much of it's actually getting recycled? Can you tell us about that? That's exciting.

Apurba Pradhan: 11:47

Yeah, this is a capability that we've been working on. We're really excited to roll this out. and we'll be rolling this out towards the later part of this year. we've been working, on sort of improving and, demonstrating this at the moment. we've gotten really good at detecting brands, and packaging, and that's not an easy task, right? once a, a package goes through all the different, equipment, at the end of that, That, line, the brand labels and things like that may or may not be visible. So there's a lot of, sort of development that we've had to do to be able to detect brands. I mean, brand brands have dozens and dozens of different packaging depending on where they're selling it. Over time these packaging evolve. So we gotta keep a good, database of what's happening with. On the packaging side of things. And then how do we track that? How do we categorize track that and detect that when it comes out, into a merf? So we've gotten really good at that. and the, there's sort of the easy things that you can do with that. And there's sort of the more not so, not so easy and not so, I guess, obvious things that we can do like that, right? Like, the easy things that we can do with that is we can obviously, help brands recover more of their packaging, right? And provide them the data aggregated at a merf, at a regional, at a sort of a national level, what we're seeing, in the merf, me that we're installed in. and to get extrapolate that data. so that's the obvious thing we can do. the. Obvious things that we can do. And I think some of the critical things that, that, the CPG brands themselves are sort of figuring out how to do is. on around education, right? So, so lot, a lot in almost every package goods manufacturer have some form of a, recyclability initiative, whether it's using more recycle content or making their packaging more recyclable, right? a lot of times it's changing the behavior, the consumer to recycle those things. That's the hard part, not the designing of those packaging, right? and oftentimes, like if you take an example of you buy coffee in a little coffee, bag, or you buy toothpaste and a toothpaste tube. Like you, you and I are not throwing that into the recycling bin. We're throwing that into the trash. so the I think that brands are sort of waking up to the fact that, hey, we need to educate people on, if we're gonna redesign packaging, we need to educate them, and at least putting it into the right bin.

Cory Connors: 14:10

that's so

Apurba Pradhan: 14:11

important. And if they're, and if they're in investing a lot of money in the packaging, they and if they want this program to succeed, they better invest, in the education part of it as well. And we can provide the data to help them understand if that education, is going well or not, right. So we can. For and after studies on how much of that is showing up on MEFs, where those packaging are actually ending up once they go into side of mef. Because that's the other thing is I talked about existing infrastructure. the existing infrastructure of MEFs is not ha equipped to handle small packaging or orange kind of nuanced packaging. And so we kind of we can also tell 'em, Hey, you're packaging showed up here and it actually ended up here in. this is so, so they can do a lot of sort of experimentation with design and, things like that. So there's a lot of sort of not so obvious information that we can provide with the data, not just sort of capturing it at the end of it, which is pretty exciting.

Cory Connors: 15:06

That sounds incredibly valuable. To, to a brand, to a packaging company maybe as well, because they're able to say, well, this isn't very recyclable. It's not easy to make it through the Merf, even with these incredible automation, automated machines. Yeah. what do you see in the future for forever Everest slaps that you're really excited?

Apurba Pradhan: 15:28

Yeah, I mean, I, our be bread and butter is, getting more and more penetration at the MEFs. so we're doing that. We're sort of quiet, quietly, sort of winning a lot of, contracts and winning a lot of business with the mfs. So that's going great. congratulations. So we're we're excited about continuing to doing that. we talk about empowering MEFs and that's kind of how we go about, working with MEFs is to provide them the data and the automation that they need, at price points that they can afford. then really the next step that's exciting for us is to be able to provide, some of this capability, to. Not just sort of the CPG companies and the brands themselves, but sort of the industry partnerships, right? right. there's industry coalitions and partnerships around various different containers and packaging types, and they're all trying to figure out what to do with the funding that they've received from there sort of, the companies that they represent and how how they can help improve recycling. And we're excited working with those types of organizations as well. companies like the Can Institute or the Container Research Institute, all these guys have money to try to figure out how to make recycling of their specific. Packaging better. Right. I don't think any of them have really figured it out because one of the things is they're all making sort of individual requests to these recycling facilities and Yeah. working with AI and robotics technologies that can sort of detect and, sort the various pa different types of packaging together in one line. Right? Can be challenged into one technology, one initiative. And then at that point it might make sense for a murf to implement something like that, right? Because it's hard for a merf to say, okay, I'm gonna sit here and collect all the different. I don't know what have you Teter pack or some kind of aerosol can, because it'll take months to fill up a bail for them to do anything with, right? But if they, all of these organizations come together and make sort of a, a. a single, sort of framework around what the MEF can do with AI and robotics technology. I think that helps move the, all of these organizations forward. So I think we're definitely excited about that. definitely excited about the brand detection and all the data that we can, capture around sort of the recyclability of different brands. And we're hoping to roll that out, over the course of the year. so yeah, a lot of exciting things happening here. But again I think this industry's, AI and robotics and recycling is still, in its early stages. We we haven't penetrated all the companies combined together haven't probably penetrated more than 10, 15% of what's possible out there. So we're still in the early days. and as we get technologies more and more into these MEFs, the data place starts becoming more and more important. Yeah.

Cory Connors: 18:12

Excellent points. I think you're exactly right. we need to understand that we're just getting started. Yeah. And oftentimes I'll get criticized for encouraging recycling, and I think that's a, it's a narrow mindset. And especially when we know what's gonna happen in the future with recycling, it's gonna get, continue to get better and better. And those materials are gonna be more and more valuable as we move forward with the onset of extended producer responsibility and the need for higher post-consumer recycled

Apurba Pradhan: 18:44

content. Yeah, that's right. I mean, I was at the Plastics Recycling Conference in DC Oh yeah. Two weeks ago, or three weeks ago. and, the general sentiment there is that high quality recycled content, the demand for that is just going up. Right. And yeah, something driven by E P R, as you state, driven by the, economics as well, right? Yeah. and as demand for these materials go up the prices go up and makes economics better. I mean, we I saw companies that were collecting high quality. plastics from the US shipping it to places like India. Yep. Processing them there and sending it and shipping it back to the US to sell, to material, packaging companies here. And there's enough money in the supply chain to do that right now. So the demand, high quality material is there. And so, I think what we're trying to do is help the most, make those quality of those materials better and,

Cory Connors: 19:42

Well keep up the good work. we're impressed with it and thankful for it. What's, what's the best way to get, in touch with you?

Apurba Pradhan: 19:49

Yeah, probably email. my, my email is my first name@everestlabs.ai. So, it's APO by P U R B A, or if you just visit Everest labs.ai, we've got some great videos and, just a lot of Eyecatching Robotics and AI in action. So take a look at that. But in there you'll also find, an email address, hello Everest Labs, and that, that comes to me as well.

Cory Connors: 20:12

Oh, nice. That's perfect. Well, thank you sir. This has been really informative and frankly exciting for somebody like me that's, that's hopeful for a future where more and more materials get recycled. So well done. Thank you for your input

Apurba Pradhan: 20:24

and your wisdom. Awesome. Thanks Corey. I appreciate it. And, thanks for having us on and, good to see you in person. You too.

Cory Connors: 20:32

If you're listening, make sure you subscribe so you don't miss the next episode. And thank you Landsberg Orora for sponsoring this show and stay tuned for more. Thank you.