Practical ways to detect plastic on a mobile robot? IR spectrum sensing?

I've made a mobile robot that can drive on sand, or really just a rover since it isn't yet automated just remote control. I was thinking it would be nice if it could find and remove plastic, we have a lot of plastic trash on our beaches. There aren't a lot of large objects and those are easier to clean up by hand. I'm thinking what requires an automated solution are the little bits. From a plastic cup down to the size of a fragment of a bottle cap.

It would be easy enough for it to sift the sand up, but it would find shells, bits of wood and pebbles in the mix. So I was looking at ways for it to detect plastic from other things (there would be some metal and glass too, but those are less problematic)

I looked up how they do it in automated recycling plants and found that they use IR laser diodes. That article was more of a news article about the service so it didn't cover the science. But doing a bit of searching on spectral plastic detection I found this:
https://www.idec.com/auen/technology_solution/our_core_tech/plastic_sensing.html

In summary, it says that plastics can be identified using their high absorption of waves near 1700 nm, using an Indium gallium arsenide (InGaAs) detector.



I know that spectral analyzers aren't cheap. But is there some simplified, task-specific way I could use this property? Thor labs have a sensor that costs $400 that could do the job. InGaAs Fixed Gain Amplified Detector, 800-1700 nm. I even found a few on Ebay for under $300. (Edit: These also just give you one measurement not a spectrum, so I guess a filter or emission sweep would be needed here as well)

I'm thinking that considering how advanced this technology is there probably isn't a simple or cheaper way. And really $400 might be doable. Even though I am doing this as a hobby project and not as a tech startup or something that would make money.

There are also devices like this InGaAs diode for $30. But it just returns a spectrum so you would either need to have a filter that changes or change your emission to find out if there was a drop at 1700 nm. I don't know if there is a feasible way to do that.

Any ideas or suggestions to this puzzle? I'm open to modifications or a totally different approach.

Thanks!

I cant do it until Tuesday, but in Stranraer near to where I live, they have a central recycle centre. The recycle stuff didnt used to take plastic there, but recently they have started after a upgrade. If you like I could go look and see what they use, there are so many different plastics, I am surprised they all absorb near the same spectrum??

A two-colour near-infrared sensor for sorting recycled plastic waste


Abstract
This paper describes a simple device for automated sorting of post-consumer plastic waste. It is shown that rapid and reliable identification of certain polymers can be achieved by using a simple two-colour fixed filter near-infrared spectrometer in conjunction with a simple ratio circuit. The effectiveness of the proposed method is demonstrated in sorting of polyethylene terephthalate and polyvinyl chloride. A sensor based on this identification method is inexpensive to build and provides the necessary speed and performance required by the recycling industry.


I erm said hi in a pm . This was all I could find HIH