A project I have been thinking about and wanting to do for a very long time is to build a fully autonomous litter collecting robot. This driven by the annoyance I always feel when passing a nearby park, littered with Twix wrappers, coke cans, and the like. Challenging? Very much so. Impossible? No. You just have to pick your constraints.
I wish this was the blog post I would explain how I built the robot and how it all works, including snazzy youtube video. However, while I have already started down the path I still have a long (but fun!) way to go. In particular my Orangutans have been keeping me busy and will continue to do so for a while. I have also changed my professional affiliation but that’s a topic for the next post. It does explain though why my interest was piqued when Peter Kohler (GIS expert from SESexplore, and Fishackathon fame) told me about his project to raise awareness around marine litter. And that is the real topic of this post.
Marine litter is a huge problem with lots of statistics that boggle the mind. It is estimated three times as much rubbish is dumped into the world’s oceans annually as the weight of fish caught. In particular this includes around 7 billion tonnes of plastic litter that enters the ocean every year. The United Nations Environment Programme estimated in 2006 that every square mile of ocean contains 46,000 pieces of floating plastic. Then consider that your average plastic bag takes 10-20 years to break down and A normal plastic bottle may persist for more than 450 years if left on a beach. Most people will also have heard of the North Pacific Gyre, or rather, the Great Pacific Garbage Patch. It occupies a relatively stationary area that is twice the size of Texas (, , , , ).
Peter’s interest in particular is to drive down the coast of West Africa mapping out beach litter patches and collecting & recycling what he can. The drive would be to raise awareness and data on litter would be recorded in a scientifically useful way such that it can be fed into the relevant models (current understanding of how litter moves around at sea is still rather crude).
To increase coverage and efficiency we figured we could employ drones to map beaches and strandlines and process the imagery with some of the fancy deep nets that are all the rage these days. Hence a plan was formed and a UK version of the drive planned late summer. To test the basic data collection workflow we bought a second hand Canon SX260 on eBay, installed CHDK and I strapped it to my venerable H500 quad. Joined by Adventurer Newall Hunter and his Insprire & Phantom we set off to a beach along the Bristol Channel to see what kind of imagery (and litter) we could collect.
The site proved to be a good choice with a staggering amount of debris and litter on the beach and the estuary behind. This despite the fact this was a conservation area with a rich bird life.
We played with CHDK settings and did a number of flights across the beach. We kept it low and slow, the main aim being to get the best imagery possible with the kit we had and cross reference what we could see on the images with what we picked on the ground.
DroneDeploy was then used to stitch together the imagery into an orthophoto:
The mosaic turned out quite blurry but anything about the size of a tennis ball or larger is clearly visible and, given the right algorithm and enough data, should be detectable. We are well aware the camera isn’t the latest and greatest but the aim was to establish a baseline.
We then proceeded to litter pick the stretch of beach and collected a trailer worth of litter in no time. We recorded what we found on an ad hoc form, and called it a day.
Lessons were learned and in order to get some more science into the project we were joined by Kayleigh Wyles from Plymouth Marine Laboratory and, from Imperial College, Michael Lange (HPC), Erik Van Sebille (Ocean circulation), and Stefan Leutenegger (Dyson lab). Some emails and a Skype call later, a new plan was made and this time Peter, Stefan, and myself headed out to a pebble beach on the Solent. The pattern was the same except things were more streamlined and more structured. Erik had kindly shared his litter survey protocol and we also started using the Marine Debris Tracker app.
We did a manual survey of a 300m stretch and mapped it with the drone. Processing was again done with DroneDeploy though this time some log file fiddling with mission planner was needed as the sx260 GPS got seriously lost.
While on the face of it this beach looked very clean indeed, we found a surprising amount of litter, particularly plastics with lots of wire insulation material, bottle tops, straws, twine, and other random bits. While not visible on the orthophoto (not quite sure what to blame for the blur) you can see the bigger bits in the original photos. The pebbles and seaweed definitely don’t make life easy though.
While we thought we would miss them it seems we also captured some sunbathers.
A Smart Litter Picker
Manual litter picking is tedious and, in some cases, depressing work. You quickly realise why the recommendation is to only pick 100-300m at a time. Hence we soon got talking about how you could develop a high tech litter picker that would automatically count and photograph anything you pick up with it and update the Litter App on your phone via bluetooth. Not quite Wall-E but definitely feasible and a wonderful project for a student or a maker with some spare time. Any takers?
With the workflow coming together we also started thinking about possible image processing techniques. Personally I have been eyeing Generative Adversarial Networks (GANs)  to help reduce the amount of labelling we’d need to do. Its still early days though and Stefan has some ideas as well. Mechanical Turk is likely to be involved at some point in any case (or perhaps we could reach out to the Zooniverse folk?). But first we need a better camera and bigger datasets, the next trip is already in the diary.
So again a wonderful little project and props to Peter for pulling it together. (Marine) litter is a massive ecological problem, an eyesore on the beach, and so easy to avoid.
Hence I really encourage everybody to get involved, be it on land or at sea. At least have a read of Erik’s recent article in the Guardian (Plastic waste dumped in UK seas ‘carried to Arctic within two years) and next time you are at a beach take a bag with you and pick up the odd bits of litter you see lying around.
For if you don’t they may well just lie there for another few hundred years.
Note that for all drone flights we worked with the local councils and conservation societies to ensure they were in the loop with what we were doing and all drone operations were conducted sensibly and safely. If you want do do some flying, please do the same.