Contest Winners: Machine Learning On All Kinds Of Gadgets

April 17, 2020 by Mike Szczys 9 Comments

With nearly sixty exciting entries, the Train All the Things contest, presented in partnership with Digi-Key, has drawn to a close and today we are happy to share news of the winning projects. The challenge at hand was to show off a project using some type of Machine Learning and there were plenty of takes on this theme displayed.

Perhaps the most impressive project is the Intelligent Bat Detector by [Tegwyn☠Twmffat] which claims the “ML on the Edge” award. His project, seen above, seeks not only to detect the presence of bats through the sounds they make during echolocation, but to identify the type of bat as well. Having been through a number of iterations, the bat detector, based on Nvidia Jetson Nano and a Raspberry Pi, can classify several types of bats, and a set of house keys (for a “control”). It’s also been impeccably documented and serves as a great example of how to get into machine learning.

The Soldering LIghtsaber takes the “ML Blinky” award for using machine learning in the microcontroller realm. This clever use of the concept seeks one thing: destroying the wait times for your soldering iron to heat up. It takes time to make temperature readings while the iron heats up, if you can do away with this step it speeds things up greatly. By sampling results of different voltages and heating times, machine learning establishes its own guidelines for how to pour electricity into the heating element without checking for feedback, and coming out the other side at the perfect temperature.

AI Powered Bulls*** Detector

Wearables Fitness Trackers for Mental Health

Rounding up our final two winners, the AI Powered Bull**** Detector claims the “ML on the Gateway” award, and
Hacking Wearables for Mental Health and More which won in the “ML on the Cloud” category.

The idea behind our illuminated poop emoji project is to detect human speech and make a judgement on whether the comment is valid, or BS. It does this by leveraging a learning set of comments that have previously been identified as BS and making an association with the currently uttered words.

Wearables for mental health is a wonderful project that was previously recognized in the 2018 Hackaday Prize. Economies of scale have made these wearables quite affordable as a way to add a sensor suite to behavior analysis. But of course you need a way to process all of the sensor data, a perfect task for a cloud-based machine learning application.

All four winners received a $100 gift code to Tindie. Don’t forget to check out all of the other interesting projects that were entered in this contest!

Hackaday Podcast 063: Magnetic Gears, AI Green Screen, Plasma <3 Sharpie, And A Rubbery Drivetrain

April 17, 2020 by Mike Szczys 4 Comments

Hackaday editors Mike Szczys and Elliot Williams sift for hacking gold from the past week. In this episode, we remember John Horton Conway’s Game of Life and its effect on novice programmers. We geek out adding screens to your car with an OBD-II hack, automating a Sharpie clicker as part of a plasma cutter, and 3D printing an incredible RC car that drives every wheel from a single motor. Plus we look at machine-learning for custom backgrounds in your video chats, take a gander at the coming generation of ePaper displays, and we get cultured about yeast.

Take a look at the links below if you want to follow along, and as always tell us what you think about this episode in the comments!

Take a look at the links below if you want to follow along, and as always, tell us what you think about this episode in the comments!

Direct download (60 MB or so.)

Continue reading “Hackaday Podcast 063: Magnetic Gears, AI Green Screen, Plasma <3 Sharpie, And A Rubbery Drivetrain” →

John Horton Conway, Creator Of Conway’s Game Of Life, Has Died

April 13, 2020 by Mike Szczys 57 Comments

Programmers everywhere are familiar with Conway’s Game of Life: whether they’ve written a version themselves or simply seen the mesmerizing action resulting from the cellular automata, it’s a household name in all homes where code is spoken. On Saturday April 11th, 2020 its inventor and namesake, John Horton Conway passed away from COVID-19 at the age of 82.

Born in Liverpool, Conway received his PhD in mathematics in 1964 from Gonville and Caius College, Cambridge. He accepted a position at Sidney Sussex College, Cambridge which he held until joining the faculty of Princeton University in 1987. A brilliant mathematician, he received numerous awards and was well known for his work in combinatorial game theory, group theory, and theoretical physics.

Many readers will be familiar with his Doomsday algorithm which can be used to deduce the day of the week for any given date in your head. But by far the rockstar mathematics moment of developing Conway’s Game of Life in 1970 cements him a perpetual place of legend in computing lore. His original work on the concept used pencil and paper as the computing revolution had yet to make digital resources easily available, even to mathematics researchers like Conway.

The game uses an infinite grid of squares where all of the edges of the grid wrap around. Four simple rules (which can be boiled down to three if you’re clever) determine which cells live and which cells die during each frame of the “game”. The only parameters that are needed are the number and position of living cells at the start of the game, and the delay between each game frame. But the effect of this simplicity is not to be understated. The game can be coded by a novice — and it’s become a common challenge in University course work. Small errors, or intentional tweaks, in the implementation have profound effects on behavior of the game. And the effect on the person programming it for the first time can be long lasting. You could call it a mathematics gateway drug, grabbing the curiosity of the unsuspecting mind and pulling it down the rabbit hole of advanced mathematics discovery.

We’d love to celebrate his life by hearing your own stories of programming the Game of Life in the comments below. If you haven’t yet had the pleasure, now’s a great time to take on the challenge.

[Game of Life example shown in this article is John Conway’s Game of Life – 1.0 written in Python by Nick Jarvis and Nick Wayne]

Clever Suction For Robot Arm Automates Face Shield Production

April 12, 2020 by Mike Szczys 9 Comments

We’re certainly familiar with vacuum grabbers used in manufacturing to pick items up, but this is a bit different. [James Wigglesworth] sent in some renders and demo video (embedded after the break) of the Dexter robot arm and a laser cutter automatically producing face shields.

It’s a nice little bit of automation, where you can see a roll of plastic on the right side of the Glowforge laser cutter feeding into the machine. Once the laser does its thing, the the robot arm reaches in and grabs the newly cut face shield and stacks it in a box neatly for future assembly. There are a lot of interesting parts here, but the fact that the vacuum grabber is doing it’s job without a vacuum air supply is the one we have our eye on.

The vacuum comes from a corrugated sleeve that makes up the suction cup on the end of the robot arm. A rubber band holds a hinged piece over a valve on that sleeve that can be opened or closed by a servo motor. When the cuff is compressed against the face shield, the servo closes the valve, using the tape as a gasket, and the corrugated nature of the cuff creates a vacuum due to the weight of the item it is lifting. This means you don’t need a vacuum source plumbed into the robot, just a wire to power the servo.

The robot arm is of course the design that won the 2018 Hackaday Prize. I comes as no surprise to see the Haddington Dynamics crew setting up a manufacturing line like this one. As we discovered a few weeks ago, 3D printers, laser cutters, and robot arms are part of their microfactory setup and well suited to making PPE to help reduce the shortage during the COVID-19 outbreak.

Continue reading “Clever Suction For Robot Arm Automates Face Shield Production” →

8mm Film Scanner Grows Into A Masterpiece

April 12, 2020 by Mike Szczys 73 Comments

Digitizing film is a tedious process that becomes a lot more fun if you spend more of your time building a digitizer and less time actually working working with old film. [Heikki Hietala] has been at it for years and his Kotokino Mark IV film scanner is a masterpiece of simple machine building.

Since we first saw the film scanner four years ago it’s undergone a number of excellent improvements. Most notably, the point-and-shoot camera has been swapped out for a DSLR. With the use of a macro reversing ring a normal lens is flipped around to blow up the 8-millimeter-wide film to take advantage of all the megapixels available on the camera sensor.

The key to the setup is the film advancer mechanism which takes care of both advancing the film and triggering the camera. As you can see, a servo motor rotating an axle provides the locomotion. The mechanism keys into the perforations in the film to pull it along on the down stroke and closes a switch to trigger the camera on the upstroke. Directly under the lens, the alignment jig uses lens cleaning fabric to avoid scratching the film, while perfectly positioning it over the light source.

Previous versions have placed the camera on the horizontal plane but it seems some vibrations in the system caused alignment problems between captured frames. This latest version places the camera pointed straight down to solve that issue, and brings the entire thing together into one beautiful finished project. Having gathered numerous fans of the build along the way, [Heikki] has made the design files available so that you may build your own version.

Hackaday Podcast 062: Tripping Batteries, Ventilator Design, Stinky Prints, And Simon Says Servos

April 10, 2020 by Mike Szczys No comments

Hackaday editors Elliot Williams and Mike Szczys check out the week’s awesome hacks. From the mundane of RC controlled TP to a comprehensive look into JTAG for Hackers, there’s something for everyone. We discuss a great guide on the smelly business of resin printing, and look at the misuse of lithium battery protection circuits. There’s a trainable servo, star-tracking space probes, and a deep dive into why bootstrapped ventilator designs are hard.

Take a look at the links below if you want to follow along, and as always tell us what you think about this episode in the comments!

Take a look at the links below if you want to follow along, and as always, tell us what you think about this episode in the comments!

Direct download (60 MB or so.)

Continue reading “Hackaday Podcast 062: Tripping Batteries, Ventilator Design, Stinky Prints, And Simon Says Servos” →

What Does A Dependable Open Source Ventilator Look Like?

April 9, 2020 by Mike Szczys 61 Comments

Ventilators are key in the treating the most dire cases of coronavirus. The exponential growth of infections, and the number of patients in respiratory distress, has outpaced the number of available ventilators. In times of crisis, everyone looks for ways they can help, and one of the ways the hardware community has responded is in work toward a ventilator design that can be rapidly manufactured to meet the need.

The difficult truth is that the complexity of ventilator features needed to treat the sickest patients makes a bootstrapped design incredibly difficult, and I believe impossible to achieve in quantity on this timeline. Still, a well-engineered and clinically approved open source ventilator might deliver many benefits beyond the current crisis. Let’s take a look at some of the efforts we’ve been seeing recently and what it would take to pull together a complete design.

Continue reading “What Does A Dependable Open Source Ventilator Look Like?” →