Karakuri Kaizen: Hacks For The Factory Floor

January 4, 2019 by 36 Comments

Anyone who has an interest and/or career in manufacturing would have heard of Kaizen, generally a concept to continuously improve your process everywhere. Under that huge umbrella is Karakuri Kaizen, encouraging workers on the factory floor to adopt a hacker mentality and improve their own work stations. It is right up our alley, manufacturer or not, making this overview by Automotive News an entertaining read.

Karakuri could be translated as “mechanism”, but implies something novel in the vein of English words gadgets, gizmos, or dare we say it: hacks. Karakuri has a history dating back to centuries-old wind-up automatons all the way to modern Rube Goldberg contraptions. When applied to modern manufacturing (as part of factory training) it encourages everyone to devise simple improvements. Each might only shave seconds off assembly time, but savings add up in due time.

Modern global manufacturing is very competitive and survival requires producing more efficiently than your competitors. While spotlights of attention may be focused on technology, automation, and construction of “alien dreadnoughts”, that focus risks neglecting gains found at a smaller and simpler scale. Kaizen means always searching for improvements, and the answer is not always more technology.

Several points in these articles asserted purely mechanical karakuri are far less expensive than automated solutions, by comparing price tags which are obviously for industrial automation equipment. We’d be curious to see if our favorite low cost tools — AVR, PIC, ESP32, and friends — would make future inroads in this area. We’ve certainly seen hacks for production at a much smaller scale.

Embedded below the break is a short video from Toyota showing off a few karakuri on their factory floor.

Continue reading “Karakuri Kaizen: Hacks For The Factory Floor” →

Pokemon Cries And How They Work

January 4, 2019 by 4 Comments

If you grew up watching the Pokémon TV series, you’d naturally be familiar with the cries of all your favourite Pocket Monsters. Most of the creatures in the anime tend to say their own name, over and over again. Pour one out for the legions of parents who, upon hearing a distant “PIKA PIKA!”,  still involuntarily twitch to this day.

However, the games differ heavily in this area. Generation I of Pokémon was released on the Game Boy, which simply didn’t have the sound capabilities to deliver full bitstream audio. Instead, sounds were synthesized for the various Pokémon based on various parameters. It’s quite a deep and involved system, but never fear – help is at hand via [Retro Game Mechanics Explained].

The video breaks down, at a bitwise level, how the parameters are stored for each Pokémon’s cry, and how they are synthesized. It’s broken down into easily understandable chunks, explaining first how the Game Boy’s sound hardware works, with two pulse channels and a noise channel, before later expanding upon why some Pokémon have the same or similar cries.

It’s a tour de force in retro game reverse engineering, and expertly presented with high quality graphical guides as to what’s going on at the software level. There’s even an emulator you can use to explore the various cries from the original game, and generate your own, too.

Now that we’re up to speed with Pokémon, how about fixing bugs in a 37 year old game? Video after the break.

Continue reading “Pokemon Cries And How They Work” →

Pi Zero Gives Amateur Astronomer Affordable Control Of Telescope

January 3, 2019 by 18 Comments

Like many other hobbies, astronomy can be pursued on many levels, with equipment costs ranging from the affordable to the – well, astronomical. Thankfully, there are lots of entry-level telescopes on the market, some that even come with mounts that automatically find and track heavenly bodies. Finding a feature is as easy as aligning to a few known stars and looking up the object in the database embedded in the remote.

Few of the affordable mounts are WiFi-accessible, though, which is a gap [Dane Gardner]’s Raspberry Pi interface for Celestron telescopes aims to fill. For the price of a $10 Pi Zero W and a little know-how, [Dane] was able to gain full control over his ‘scope. His instrument is a Celestron NexStar, a Schmidt-Cassegrain reflector with a 150-mm aperture, has a motorized altitude-azimuth mount. The handheld remote had enough room for him to add the Zero, powering it from the mount’s battery pack. The handset has an RS-232 serial port built-in, but with the level differences [Dane] just connected the Pi directly to the handset before the UART. Running INDI, a cross-platform astronomical instrument control library, he now has total control of the scope, and he can use open source astronomy software rather than the limited database within the handset. As a neat side trick, the telescope can now be controlled with a Bluetooth gamepad.

Astronomy and electronics go hand in hand, whether in the optical or radio part of the spectrum. We like the way [Dane] was able to gain control of his telescope, and we’d like to hear about what he sees with his new tool. Assuming the Seattle weather ever cooperates.

Continue reading “Pi Zero Gives Amateur Astronomer Affordable Control Of Telescope” →

Cheating AI Caught Hiding Data Using Steganography

January 3, 2019 by 63 Comments

AI today is like a super fast kid going through school whose teachers need to be smarter than if not as quick. In an astonishing turn of events, a (satelite)image-to-(map)image conversion algorithm was found hiding a cheat-sheet of sorts while generating maps to appear as it if had ‘learned’ do the opposite effectively[PDF].

The CycleGAN is a network that excels at learning how to map image transformations such as converting any old photo into one that looks like a Van Gogh or Picasso. Another example would be to be able to take the image of a horse and add stripes to make it look like a zebra. The CycleGAN once trained can do the reverse as well, such as an example of taking a map and convert it into a satellite image. There are a number of ways this can be very useful but it was in this task that an experiment at Google went wrong.

A mapping system started to perform too well and it was found that the system was not only able to regenerate images from maps but also add details like exhaust vents and skylights that would be impossible to predict from just a map. Upon inspection, it was found that the algorithm had learned to satisfy its learning parameters by hiding the image data into the generated map. This was invisible to the naked eye since the data was in the form of small color changes that would only be detected by a machine. How cool is that?!

This is similar to something called an ‘Adversarial Attack‘ where tiny amounts of hidden data in an image or other data-set will cause an AI to produce erroneous output. Small numbers of pixels could cause an AI to interpret a Panda as a Gibbon or the ocean as an open highway. Fortunately there are strategies to thwart such attacks but nothing is perfect.

You can do a lot with AI, such as reliably detecting objects on a Raspberry Pi, but with Facial Recognition possibly violating privacy some techniques to fool AI might actually come in handy.

A Low Cost VR Headset

January 3, 2019 by 3 Comments

Virtual reality systems have been at the forefront of development for several decades. While there are  commercial offerings now, it’s interesting to go back in time to when the systems were much more limited. [Colin Ord] recently completed his own VR system, modeled on available systems from 20-30 years ago, which gives us a look inside what those systems would have been like, as well as being built for a very low cost using today’s technology.

The core of this project is a head tracker, which uses two BBC Microbits as they have both the accelerometer and compass needed to achieve the project goals. It is also capable of tracking an item and its position in the virtual space. For this project, [Colin] built everything himself including the electronics and the programming. It also makes use of Google Cardboard to hold the screen, lenses, and sensors all in the headset. All of this keeps the costs down, unlike similar systems when they were first unveiled years ago.

The ground-up approach that this project takes is indeed commendable. Hopefully we can see the code released, and others can build upon this excellent work. You could even use it to take a virtual reality cycling tour of the UK.

Continue reading “A Low Cost VR Headset” →

The Solderdoodle Open Source Iron Rides Again

January 3, 2019 by 25 Comments

Over the last year or so, cordless portable soldering irons have become all the rage. In fact, at this point a good number of Hackaday readers out there have likely traded in their full-size AC irons for a DC iron that’s only slightly larger than a pen. But before the big boom in portable irons, in the ye olden days of 2014, we brought you word of the open source Solderdoodle created by [Isaac Porras]. Based upon the Weller BP645 and featuring a 3D printed case, the DIY iron was designed to be charged from a standard USB port.

Now, [Isaac] is back with an updated version he calls the Solderdoodle Plus. It’s still based on the heating element from the Weller BP645, but now boasts twice the power, an improved 3D printed case, an intuitive touch-based user interface, and even some LED blinkenlights for good measure. As with the original Solderdoodle the hardware and software for the device are open source and you’re invited to build your own, though kits are also available through an already fully-funded Kickstarter campaign.

[Isaac] says that the temperature control functions on traditional corded soldering irons waste energy due to the large thermal mass they have to bring up to temperature. But with less thermal mass and a system of variable duty cycle pulsed power, he says the Solderdoodle Plus can do the same work as an old-school 60 watt iron while only consuming 10 watts. This allows the iron to maintain a constant 500°C for over an hour on the dual internal Panasonic NCR18500A lithium-ion batteries, and means you can charge it up with nothing more exotic than a micro USB cable.

Continue reading “The Solderdoodle Open Source Iron Rides Again” →

Reset Your Router The Modern Way

January 3, 2019 by 35 Comments

Many Hackaday readers will be settling back into their lives after a holiday period crammed into some family matriarch’s house along with too many assorted relatives, having given up their speedy internet connection for whatever passes for broadband wherever Granny lives. The bargain-basement router supplied by the telephone company will have spent the period wilting under the pressure of a hoard of teenagers watching other teenagers inanities on YouTube, and the Christmas ritual of Resetting The Router will have been performed multiple times.

A very simple schematic for the resetter.
A very simple schematic for the resetter.

Wouldn’t it be nice if your router simply reset itself every time it crashed or the Internet connection went down? [Cyb3rn0id] has a solution (Italian original here), in the form of an ESP8266 that pings an online service every few seconds, and turns the router off and on again via a power relay in the event that the ping attempt is repeatedly unsuccessful. It’s brilliantly simple, requiring only a single GPIO and a MOSFET to fire the relay with an LED indicator for good measure, and it’s built upon a piece of prototyping board. The router power is switched on the low-voltage side for safety.

The software is pretty basic and has the WiFi credentials hard-coded into it, so we’re guessing a version with a web interface could be built. But as a personal device for easing the pain of router crashes it gets our vote despite that shortcoming.

This isn’t the first router resetter we have seen here, but a previous model still required human intervention.