Laser to detect plastic/alloy materials

[Elopement]

Walking along the road our way in the countryside I notice food snack wrappers on the roadside all of the time: water bottles (glass and plastic), cans, crisp packets, the usual. Instead of manually picking them up and focusing my eyes on the floor, instead of the scenery!

I would like a arial drone to scout the area beforehand to catalogue each visible item, which I believe lasers have been used to determine materials in factory conditions.

The laser would only need to be able to discern between plastic or alloy from natural dirt and foliage found on the roadside (of a certain size (i.g. ignore fence gates made of steel)), and preferably on a single pass.

What laser would I need to achieve this please and is there one commercially available off of the shelf?

 

[paul1598419]

I am unaware of any laser technology that can do this. It could be done with cameras and the right software.

 

[ragebot]

Not sure why you would want to use a laser for this application. First off lasers at one level can be viewed as focused light which means the area the photons hit is limited. Even a very diffused laser beam would be fairly small in terms of area, so multiple passes would be required for even a small area.

On the other there are many thermal cameras that could detect temperature differences between man made and natural materials. For fairly small cost (less than $US10,000) a basic fixed wing RC controlled platform with a decent thermal camera can be bought and entry level ones are even cheaper. Not to mention there are lots of companies/individuals who provide thermal imaging services for things like agriculture, solar energy installations, home and industry roof inspections, and power company inspections to name just a few. Many of them have platforms that can fly a programed flight course over fields, electric transmission lines, or solar panel fields and send the thermal images real time to a ground station as well as provide large scale panos of thousands of images once the flight is completed. A quick google search on the term UAV infrared will turn up literally millions of hits. Here is a link to get you started.

 

[Elopement]

Thanks for the replies.

I am sure I read somewhere that there was a recycling plant which could use lasers to detect the type of material and automatically sort the plastic accordingly:

https://www.acreo.se/projects/laser-methods-for-automatic-scrap-metal-sorting
https://www.popsci.com/technology/a...-recycling-machines-separate-junk-type#page-2
https://waste-management-world.com/a/laser-gets-it-sorted-a-revolutionary-sorting-machine


I was thinking of spraying the beam out like in a disco, it sweeping an area in a single pass, rather than a concentrated beam dot.

 

[Benm]

Yes, lasers are used in spectroscopy for plastic recycling, basically telling one kind of plastic from another allowing you to sort shredded waste material.

For example it could identify PET (soda bottle shreds) and sort that into a bin for recycling, PP (packaging material) into another etc.

This method works mostly at a pretty close range (under a meter), but more importantly only sorts material that is known to be trash.

If you could make a trash collecting drone based on the technology, how do you prevent it from taking things that are not trash but intentionally there? If the top of my solar garden lights is PET, should it keep stealing those from me? Or what about something like plastic traffic cones, that are probably there for some good reason?

Maybe it could work to clean up a beach or something where there is not supposed to be any of these materials present (and/or you can map the ones that are in permanent locations and desirable such as plastic posts, signs etc). As long as you do your cleanup run when there are absolutely no people around that might work. If there are, some will probably get very angry when your drone steals their soda bottle

 

[Cyparagon]

Paul's got it. Camera and software is the cheapest, easiest, simplest, and safest way to accomplish something like that. It would still be wildly complex software requiring a team of engineers. Unless you've got a million dollars to invest in this project, abandon ship now.

I am sure I read somewhere that there was a recycling plant which could use lasers to detect the type of material and automatically sort the plastic accordingly

That only works because the system expects all input materials to fit into a small number of categories. Mount a simplistic system like that on a drone, and it is NOT going to know what to do with vehicles, puddles, pylons, or pedestrians for example.

 

[Zom-B]

You can use x-ray (is there a laser for that?) to determine materials probably in a Raman-style way and looking at fluorescence. It requires a (close to) dark room. For example, the scanner Cody's Lab uses: https://youtu.be/a_EdGQyHTxU?t=56s

There's no way to do that 'at a distance', let alone flying distance.

 

[Elopement]

I was interested in the drone scanning along the roadside only, around my area its pretty clean cut what is natural matter and what is human waste.

I will investigate the photography option, best start simple as you point out.

Yes, lasers are used in spectroscopy for plastic recycling, basically telling one kind of plastic from another allowing you to sort shredded waste material.

For example it could identify PET (soda bottle shreds) and sort that into a bin for recycling, PP (packaging material) into another etc.

This method works mostly at a pretty close range (under a meter), but more importantly only sorts material that is known to be trash.

If you could make a trash collecting drone based on the technology, how do you prevent it from taking things that are not trash but intentionally there? If the top of my solar garden lights is PET, should it keep stealing those from me? Or what about something like plastic traffic cones, that are probably there for some good reason?

Maybe it could work to clean up a beach or something where there is not supposed to be any of these materials present (and/or you can map the ones that are in permanent locations and desirable such as plastic posts, signs etc). As long as you do your cleanup run when there are absolutely no people around that might work. If there are, some will probably get very angry when your drone steals their soda bottle

 

[kecked]

Lidar might do it but expensive. Terahetz waves will do it but not at a reasonable distance. Students feed snacks would be cheapest.

 

[Cyparagon]

pretty clean cut what is natural matter and what is human waste.

No! It's clear to YOU what is trash or not. It is QUITE another task trying to explain to a computer how to identify things.

Still don't understand? Captchas are a good illustration of this concept. It's clear to YOU what this image says. Even a child or a foreigner could re-type these words, but it is very difficult to get an algorithm to read it. This is how many sites will differentiate between a person and a bot. If programmers really struggle with two words on a screen, what gives you the idea that trash in a ditch is any easier?

 

[Benm]

To be honest that type of captcha is -very- easy to solve by a computer, this example is so simple that basic OCR software will probably get it right. This is also why recaptacha abandoned this type a year or so ago, they became so easy to machine-solve that they were no longer useful.

They actually tried a lot of variations of it with lines drawn over the letters so algorithms could not separate the letters that easily. Then they resorted to inverting black/white in certain parts of that, all to little result eventually.

You still see a lot of similar captcha solutions out there though, and there work because of one simple reason: If you are auto-submitting forms, taking the easiest ones is the fastest way to get results. So even a simple to solve captcha is not worth the effort and spammers proceed to a form without any.



But back to the question of what is trash, and what is not: Doing so by just optical imaging could still be viable just from the drone camera footage. I'm sure you could train an AI to some degree to actually see what is trash if you construct a learning set. This would probably involve taking thousands of areal photographs and manually identifying trash to create the learning set, but that's doable.

Subsequently you could build in a system that checks if something has been there for some time. For example, if it was in the same place yesterday. That would probably prevent you from snatching my beer can i put in the sand while drinking it from one that got left behind.

I'd be interesting to see the resulting system as i know a lot of (mostly beach) places that have heaps of trash on them. Perhaps those would be a good target to start at since the trash is usually left at the flood line, so you know where to look.

You could do a first scan at low tide, and then another, say, a hour after high tide and look at what's different. Anything that's added is -probably- trash, though some of it could be fairly harmless stuff like driftwood, shells, coconut husks etc. It may still require some identification though, you wouldn't want to pick up some hermit crab in it's shell and give it a drone-lift to the incinerator.

 

[Elopement]

I need further consultation on this, I am still wedded to the idea of a laser based system because it provides indisputable feedback, where as an image needs to be trained, a laser will know.

As the system will not physically collect refuse, merely scout and collect location data, whether it recognises living animals or not is not important, as photography attached to the location would be viewed and sorted by a human, as they would be the workforce sent to deal with the refuse.

However, in a follow up model it would likely be a single-pass system that did collect refuse, and any animal would like run away on such attempt(s), but lasers can detect living objects.

 

[kecked]

I'm going to sound like a jerk in this post and I apologize it is not my intention.

A long time ago I was taught that the more complex we make something the less likely it is to be an optimized response. It is also known as the KISS principle. Keep IT Simple Stupid or Keep it Stupid Simple your choice. I try to live by this.

By the time you find and track and log and then give it to the people to go get it they could have walked or driven the path and picked it up. I think this is a great idea without a purpose. I the drone could find it and get it that makes sense but if people still have to go there then this is not a viable solution.

 

[Cyparagon]

a laser will know.

Please explain how.

 

[Elopement]

kecked: I plan to keep it simple by simply collecting data to begin with, then modify a version afterwards for collection. The idea is relieve humans of the task of collecting litter.
Cyparagon: Spectroscopy.

 

[Cyparagon]

There are countless ways to employ spectroscopy and many types of spectroscopy.

Me: "How are you going to get to your destination"
You: "Transportation"

You've clarified nothing.