Man Buys Two Metric Tons of LEGO Bricks; Sorts Them Via Machine Learning

iStock // Ekaterina Minaeva
iStock // Ekaterina Minaeva

Jacques Mattheij made a small, but awesome, mistake. He went on eBay one evening and bid on a bunch of bulk LEGO brick auctions, then went to sleep. Upon waking, he discovered that he was the high bidder on many, and was now the proud owner of two tons of LEGO bricks. (This is about 4400 pounds.) He wrote, "[L]esson 1: if you win almost all bids you are bidding too high."

Mattheij had noticed that bulk, unsorted bricks sell for something like €10/kilogram, whereas sets are roughly €40/kg and rare parts go for up to €100/kg. Much of the value of the bricks is in their sorting. If he could reduce the entropy of these bins of unsorted bricks, he could make a tidy profit. While many people do this work by hand, the problem is enormous—just the kind of challenge for a computer. Mattheij writes:

There are 38000+ shapes and there are 100+ possible shades of color (you can roughly tell how old someone is by asking them what lego colors they remember from their youth).

In the following months, Mattheij built a proof-of-concept sorting system using, of course, LEGO. He broke the problem down into a series of sub-problems (including "feeding LEGO reliably from a hopper is surprisingly hard," one of those facts of nature that will stymie even the best system design). After tinkering with the prototype at length, he expanded the system to a surprisingly complex system of conveyer belts (powered by a home treadmill), various pieces of cabinetry, and "copious quantities of crazy glue."

Here's a video showing the current system running at low speed:

The key part of the system was running the bricks past a camera paired with a computer running a neural net-based image classifier. That allows the computer (when sufficiently trained on brick images) to recognize bricks and thus categorize them by color, shape, or other parameters. Remember that as bricks pass by, they can be in any orientation, can be dirty, can even be stuck to other pieces. So having a flexible software system is key to recognizing—in a fraction of a second—what a given brick is, in order to sort it out. When a match is found, a jet of compressed air pops the piece off the conveyer belt and into a waiting bin.

After much experimentation, Mattheij rewrote the software (several times in fact) to accomplish a variety of basic tasks. At its core, the system takes images from a webcam and feeds them to a neural network to do the classification. Of course, the neural net needs to be "trained" by showing it lots of images, and telling it what those images represent. Mattheij's breakthrough was allowing the machine to effectively train itself, with guidance: Running pieces through allows the system to take its own photos, make a guess, and build on that guess. As long as Mattheij corrects the incorrect guesses, he ends up with a decent (and self-reinforcing) corpus of training data. As the machine continues running, it can rack up more training, allowing it to recognize a broad variety of pieces on the fly.

Here's another video, focusing on how the pieces move on conveyer belts (running at slow speed so puny humans can follow). You can also see the air jets in action:

In an email interview, Mattheij told Mental Floss that the system currently sorts LEGO bricks into more than 50 categories. It can also be run in a color-sorting mode to bin the parts across 12 color groups. (Thus at present you'd likely do a two-pass sort on the bricks: once for shape, then a separate pass for color.) He continues to refine the system, with a focus on making its recognition abilities faster. At some point down the line, he plans to make the software portion open source. You're on your own as far as building conveyer belts, bins, and so forth.

Check out Mattheij's writeup in two parts for more information. It starts with an overview of the story, followed up with a deep dive on the software. He's also tweeting about the project (among other things). And if you look around a bit, you'll find bulk LEGO brick auctions online—it's definitely a thing!

Ninja’s Hot & Cold Brewed System Is the Only Coffee Maker You’ll Ever Need

Amazon
Amazon

Update: The Ninja Hot & Cold Brewed System is on sale for $120 ($40 off) for Sam's Club members until May 19.

For people who just want a cup of joe to help them get out the door in the morning, the French presses, Chemexes, Aeropresses, Moka pots, and other specialized devices that coffee aficionados swear by probably seem more overwhelming than appealing. Ditto the fancy cappuccino machines at local cafes. That’s where Ninja’s new Hot & Cold Brewed System comes in: It was created to give coffee addicts a myriad of options with minimal fuss, not to mention minimal equipment. And it makes tea, too!

“Coffeehouses are known for having an endless selection, but current at-home brewers haven't given users the vast variety of choice we thought possible, and certainly not all in one product," Mark Rosenzweig, CEO of SharkNinja, said in a press release. "The Ninja Hot & Cold Brewed System changes the category entirely. This innovative system is more than just a machine you use in the morning; it's your all-day brewing partner.”

The Hot & Cold Brewed System comes with two baskets: one for coffee and one for tea. It knows what you're making to make based on the basket you insert, and the available options for that basket will light up. The machine allows the user to make six different sizes of coffee or tea, from a single cup all the way up to a full 50-ounce (10-cup) carafe.

And of course, as the name suggests, the system can make both hot and iced beverages. For coffee, it has five brew options: classic, rich, over ice, cold brew, and specialty (a concentrated brew for milky drinks like cappuccinos). If you’re making tea, you can choose between hot and cold brews optimized for herbal, black, oolong, white, or green tea.

When you select an over ice or cold brew, the machine automatically doubles the strength of your beverage so it doesn't get overly diluted by the ice cubes in the carafe. Even better, the Ninja can make cold brew in just 10 to 15 minutes, whereas other systems and methods typically take hours. (Hot coffee is brewed at 205°F, while the cold brew is made at 101°F.) And the system has a hot and cold frother that folds into the side so you can make barista-level lattes, too.

These bells and whistles sound impressive on paper, but how do they perform in real life? Ninja sent me Hot & Cold Brewed System to test for myself.

Ease of Use

Though it might look like something developed by NASA, the Hot & Cold Brewed System is designed to easily work with the twist of a dial and the push of a button, and it delivers. From loading in the correct amount of grounds with the system’s “smart scoop” to picking what type of brew you’d like, it’s simple enough to use even while bleary-eyed in the morning. It’s also easy to schedule a delayed brew so you can do the rest of your morning routine while your coffee brews. (Here’s the only drawback I can think of about this machine: When it starts brewing, it’s kind of noisy—loud enough to make my cats jump. It’s not a dealbreaker, but if you live in a small apartment and plan to brew coffee so that it’s ready right when you wake up, it might be something to consider.)

The system even tells you when it needs to be descaled. The “clean” button will light up, at which point you simply fill the water reservoir with descaling solution and water and press the clean button. A countdown lets you know how much longer the clean cycle will last.

Taste and Flavor

I swapped out an old, cheap coffee maker for the Hot & Cold Brewed System, and the difference was immediately noticeable. Whether hot or cold, the coffee made by the H&CBS was a better, smoother cup of joe. That’s due to what Ninja has dubbed Thermal Flavor Extraction automated brewing technology, which, according to a press release, “knows the precise temperatures, correct bloom times, and proper levels of saturation for every possible beverage combination to ensure a great taste every time.”

Whatever tech they use, it works. The coffee I make in this machine is consistently tasty. The rich brew setting works exactly as advertised, too, providing a richer, bolder flavor than the classic brew.

Features and Accessories

One of the best things about the H&CBS is the fact that it cuts down on waste significantly. Unlike other machines, it doesn't require any plastic pods or paper filters. Instead, it comes with two permanent filters, one for coffee and one for tea.

And the cold brew function is a game changer if you prefer iced coffee to hot. Not only does it brew quickly, but it eliminates the messy cleanup that comes with making cold brew yourself.

Typically priced at $230 for the thermal carafe version (or $200 for the glass carafe), the Hot & Cold Brewed System is significantly more expensive than a simpler drip coffee machine. But if you’re a cold brew addict looking to treat yourself, it’s worth it. Consider springing for the slightly more expensive thermal carafe model, which will keep your java hot or cold for hours. (I’ve left ice in it overnight and found cubes the next morning.)

You can get the Hot & Cold Brewed System on Amazon, Walmart, Macy's, Sam's Club, or directly on Ninja’s website starting at $160.

Springer Nature Has Published the First AI-Written Textbook

iStock.com/PhonlamaiPhoto
iStock.com/PhonlamaiPhoto

The first AI-written textbook is here, and its tech-heavy subject is exactly what you might expect from a machine-learning algorithm. As Smithsonian reports, the book, published by Springer Nature, is a 247-page guide titled Lithium-Ion Batteries: A Machine-Generated Summary of Current Research.

While it doesn’t exactly make for light reading, the fact that it was written entirely by Beta Writer—an algorithm designed by researchers in Germany—is a game changer. Sure, AI has dabbled with writing before, helping journalists pen articles and even crafting entire chapters for the Game of Thrones and Harry Potter series. (We highly recommend the riveting tale of Harry Potter and the Portrait of What Looked Like a Large Pile of Ash.) But this is the first time AI has authored an entire research book, complete with a table of contents, introductions, and linked references.

The information was pulled from Springer Nature’s online database. While the grammar and syntax are a little clunky, the book manages to get the point across. (Here’s one sample sentence: “Respectively, safety issue is apparently challengeable till now even after the first commercialization of lithium-ion battery.”)

With the exception of an introduction to the book that was written by Henning Schoenenberger, Springer Nature's director of product data and metadata, the finished product was left unedited and unpolished. This was done “to highlight the current status and remaining boundaries of machine-generated content,” according to Schoenenberger. The publisher hopes to experiment with AI-powered textbooks on other subjects in the future.

Artificial intelligence has certainly come a long way in recent years, and algorithms have been trained to carry out a number of oddly specific tasks. They can design beer, figure out the ingredients in your meal, find Waldo in a “Where’s Waldo” picture, and remake the music video of “Total Eclipse of the Heart.” In one of the more meta developments in tech news, Google’s AI even learned to make its own AI in 2017.

[h/t Smithsonian]

SECTIONS

arrow
LIVE SMARTER