Industrial AI - Part 6: A *Shiny* Idea, Whilst at Lion Vision: “Hi Vis Batteries”, and Why You Need Underdog Engineers.

A *Shiny* Idea, Whilst at Lion Vision: “Hi Vis Batteries”

I was working late with George one evening on the Sweet Detection demo for this blog, and after a long day, we headed to get some dinner. I was absent-mindedly watching the reflection of reflective strips on cyclists, road signs, traffic cones, backs of vans, security personnel, and road markings - and realising how it reflected our car’s headlamp light back at us with dazzling efficiency.

Above: Retroreflectors: 1. ‘Corner’ or 2. Spherical. Both reflect light from the same direction it came from. Image Credits: Wikipedia. Inspired by cat’s eyes, retroreflectors are even on the Moon, because science!

I wondered if we coated Li-Po / Li-Ion battery cells in this stuff, and reduced the aperture of the camera (making it darker / less light getting to the sensor) and only this ‘hi-vis’ stuff would ‘pop’ in a bright contrast.

Above: sources of inspiration on the ride home….watching hi-vis objects ‘pop!’ in our headlights.

The conventional use of the camera is to have it ‘seeing what a human sees’, but instead, I wanted it to be ‘dimmed down’ to see very little apart from the dazzling glare of the hi-vis material. The reason for this was that to the ‘human eye’ a hi-vis strip and a piece of aluminium look pretty much the same in ambient [multidirectional] light, as shown below…

Above: Ambient multi-directional light does not bring out the best reaction, as the camera feed hardly notices the difference. Remember headlights are a ‘beam’, so we need to recreate that in the setup.

It’s only when you ‘starve’ the camera sensor of light, that you only see the high contrast of the reflective tape, as shown here…you now can’t see the ‘silver’ of the aluminium trash and [conventional] aluminium batteries, as they fade to murky black. The jacket’s strips on the other hand ‘pop’ with light!

Above: General mess from eWaste processing (smashed up appliances and batteries) - next to a Hi-Vis Jacket - note this is in ambient light, all looks ‘greyish-silver’. Second image, with camera aperture closed down, and a small light shining on the subject - the Hi-Vis now ‘popping’ with high contrast!

I purchased some ‘hi-vis cyclist’s tape’ and got to work the next day - sticking it to the Lithium Batteries to see how they looked. As said in the video - I was ‘grinning like an idiot’ with excitement, as how this could be a game-changer for detection of unwanted batteries in a recycling plant. Below shows one ‘hi-vis’ battery, surrounded by 7 others - and this was only using Scott the cameraman’s spare spotlight suspended from the ‘triangular’ Lion Vision frame. We hadn't even purchased a ‘serious’ high power ‘ring light’ to go around the camera, and it was already looking very promising!

Above: A ‘Eureka Moment’ of seeing the theory work in practice. 8 batteries, with one (centre) being coated in retroreflective tape - and being far more identifiable because of it!

Playing “Where’s Wally?” with Batteries in eWaste.

AI has been tasked to ‘find’ Wally, the legendary ‘hidden’ cartoon character. This is of course a piece of satire/art - as it defeats the whole point of taking pleasure in finding something, as a game.

For George and I, finding batteries in eWaste is only slightly less fun than a game of Where’s Wally?, and yet the approach is pretty similar - but the magnitude of the task is of course colossal, as Lion Vision processes around 100 tons of waste electrical and electronic equipment (WEEE) daily, detecting approximately 4,500-cylinder batteries each day! So once again, this would perhaps get dull after a while, and better for a machine to get involved.

Here are some more tests by George at Lion Vision, further refining the concept / demonstration…

Above: Where’s Wally? meets AI. Lion Vision’s camera, with adjustable aperture settings, to let more light in (see image 3) and less light (image 4) as desired. Image 4 showing the aluminum, scrap metal, shiny plastic, etc. - and - LiPo batteries all ‘blend in’ in ambient and very low light.

Above: same setup, but with Reflective Tape on LiPo Batteries, (in the same positions). The LiPo Batteries now clearly ‘pop’ in contrast to the blurry darkness of everything else that is not as ultra-reflective (retroreflective) as the special tape. (The last image has been ‘over-exposed’ to show it’s still the same pile of eWaste and no trickery… It really works!)

What Next?

I have given many presentations to students, companies and institutions to “Trust The Process”, as others have told me, in a long tradition of ‘creatives’ having to reassure themselves that on Day 1, things will feel ‘unremarkable’, but if they remain curious, open-minded, and persistent - something usually ‘turns up’ that is worth a second look. Given that I’d gone out on a limb a little with this pitch, I was especially pleased that after all the preparation, we discovered this ‘bonus’ of not just demonstrating the work of Lion Vision; the Sweet Demo working; and expanding the Fight to Repair narrative - but potentially having a useful idea to contribute to future/pending legislation.

Above: A crude bit of ‘ball-park’ research on Alibaba to check if wholesale Aluminum Foil vs Reflective Material are wildly different in price, and it appears it’s close enough to consider further investigation.

Without wishing to get carried away, I’m aware that even if some EU legislator happens to like this idea, there are many hurdles, not least cost. Although a crude and cursory look at wholesale prices of PE Laminate (what batteries are wrapped in presently) - vs - Reflective Material (Generic), the prices are at least in the same ‘ball-park’. Although much validation would be needed, this is not a ‘laugh you out of the room’ price difference of 100x or 1000x. Although as one of my contacts, I discussed this with said, ‘if legislation says its law - it is - regardless of [modest increases] cost’.

Plan B - Blue Food Plasters // Hot Pink Batteries // Iconic Battery Design.

If you’ve ever worked in a professional kitchen, and cut your finger, you’ll notice that ALL the plasters are Blue. This is because the ‘beige’ or even ‘brown’ plasters can fall off and get into food and because many foods are these colours, nobody notices until it’s too late, and you get sued! There are virtually no ‘truly blue’ foods, as even ‘blueberries’ are not this colour.

Oddly Li-Ion Batteries seem to have a wide variety of colours. This is perhaps helpful for identification to a given manufacturer, as some use colours to indicate capacity, even if the size is the same. However, there seems to be no universal standardisation of this. I certainly have about 6 different types of Li-Ion and LiPo batteries, all in the same ‘blue’ colour tone, with very different characteristics. However, the issue is that Blue does not ‘pop’, as plenty of bits of plastic and such are blue also.

I would propose Hot Pink or Magenta to colour internal batteries (if the Hi-Vis option was not viable), as at least this way, companies like Lion Vision would be able to spot a vivid colour.

Interestingly, George told me that standard household alkaline ‘Duracell’ batteries are the easiest to ‘detect’ - as they have an iconic combination of Black and Gold, with White text. This is worth considering that whatever the Battery Standards do become, they should take note of this - and make a global design schemes that means they are all as ‘visible as possible’ to a Machine Learning Camera, like that of Lion Visions, and to consult on what are the most differentiated finishes to avoid confusion with non-batteries.

It should also be said that Hi-Vis tape comes in pretty much any colour you want, and could likely come in Magenta also. Perhaps if cost were permissible, we’d have a ‘Duracell-like’ iconic design, (striped, criss-crossed, whatever, but with that ‘head-and-body’ ratio), but that the colour was also Hi-Vis.

Legislation of Lithium Batteries - It Needs a Redesign and a Rethink.

Who’s to say where this will go, but it’s exciting to see if this can be proposed to those who review such legislation to see if a cost-effective, sustainable, and perhaps even just well-designed systematisation of Lithium batteries is just overdue and a good idea, even if totally different from the above suggestions. I’m not planning on registering any design or patenting this, as it would be rewarding enough to be acknowledged as starting the right discussion that leads to a common-sense update to this.

If you happen to know anyone working in these niche specialisms, who might be willing to help, do get in touch!

Until then I hope you’ve enjoyed this creative ‘sidecar’ to the main ‘vehicle’ that is Machine Learning and Right to Repair / eWaste awareness. I hope you’ll similarly reconsider the foundations of your own industry, as perhaps like George and I, you’ll realise something doesn’t make sense, and it’s worth changing, as well as implementing the tech as well.

Side Note: Underdog Engineers, and Why They Help You Start With Machine Learning.

I’ve been an Engineer, and Tech Scout for the likes of Dyson, and LEGO, respectively - and although I’ve seen some of the fanciest ‘foosball table offices’ around the world, complete with in-house coffee baristas, all of which certainly have their particular allure, I’ve also come to develop a genuine love for the un-showy, unpretentious, and basically ‘down-to-earth’ labs and lockers where nonetheless some impressive tech gets done, without a Flat White coffee in sight.

Indeed, whilst writing this article, I’ve been reading the Nobel-Prize-winning Richard Feynman’s autobiography: He has an even more outspoken disdain for those who flex or pontificate, without actually getting their hands-dirty, and developing an intuitive tacit knowledge of how things work. Although it’s important to not romanticise the ‘lone genius myth’, one mustn’t ignore that the ‘underdog(s)’ are often underestimated in contrast to more ‘flashy’ counterparts riding the wave of hype, (but also many succumbing to its fickle hubris). “Fast up; fast down”, as some more seasoned VCs warn. For me, Lion Vision has a more humble but stable approach - it’s not claiming to be the next ‘Unicorn’, but is just steadfastly getting on with the task at hand, and excelling in a niche which is certainly only going to grow.

This is exactly where I have seen great technology emerge that gets overlooked - where you have that ‘trinity’ of i) being in a cheaper city, than say London. ii) being next to a University advancing AI/ML techniques, and iii) being next door to an established industry (Benson’s exhausts and automotive spares). George Hawkins, a couple of years out of university, has all these hallmarks of being the ‘Young Blood’ helping usher in the new capability around Machine Learning. His boss, Jonathan Bowl, although older, is no ‘Old Guard’, and is empowering George to experiment and take chances in how to approach a novel and complex problem with purpose: Lion Vision are tackling a serious problem, and as they say ‘there’s money in muck’, and given the mess the world is in with eWaste - I’m glad they are pioneering methods which not only remove the risk of fire, but also help companies become more sustainable - perhaps even circular - in reclamation of precious minerals and metals which might otherwise sadly become pollutants in landfill.

Having spoken with 3 cohorts of design and engineering graduates in 2024, (and many in previous years) I can say with confidence that the ‘talent’ coming into their first career are indeed ‘voting with their feet’ and are forgoing some of the ‘big bucks’ to be another ‘cog in the machine’, in favour of more impactful and purpose-led work like this, where one can rise up the ranks, developing products that have a higher percentage of one’s direct input clearly visible. Indeed, George is not just writing the code, but also working with clients. He is installing the right camera and hardware for the job, and learning how to adapt it with every new engagement. If it breaks, he fixes it and improves it. Having worked in a similar capacity at points in my career, for some this is more valuable than designing a tiny component of another smartphone or other such device which frankly would have shipped without your input. George is not just the ‘youngblood’, he, along with the rest of the team - are the ‘lifeblood’.

Industrial AI Blog Series Contents:

Part 1: Lion Vision, AI vs Automation, and Why a Game of ‘Go’ Changed Everything.

Part 2: “Dirty, Dangerous, Difficult & Dull” - The Case for Ethical AI Automation.

Part 3: Key ML Terminology: Are You 'Sorting Ingredients' or 'Baking Cakes'?

Part 4: ML Lessons from Lion Vision. AI Failures, and ‘Sensing Like A Robot’.

Part 5: Getting Started with Jetson Nano / Orin. And Why Octopus Brains ML Marvels.

Part 6: A *Shiny* Idea, Whilst at Lion Vision: “Hi Vis Batteries”. And Why You Need Underdog Engineers.

Are you looking for additional information on Lion Vision?

Request more info