Self-Learning Helicopter Uses Neural Network

April 22, 2014 by Will Sweatman 24 Comments

Though this project uses an RC helicopter, it’s merely a vessel to demonstrate a fascinating machine learning algorithm developed by two Cornell students – [Akshay] and [Sergio]. The learning environment is set up with the helicopter at its center, attached to a boom. The boom restricts the helicopter’s movement down to one degree of motion, so that it can only move up from the ground (not side to side or front to back).

The goal is for the helicopter to teach itself how to get to a specific height in the quickest amount of time. A handful of IR sensors are used to tell the Atmega644 how high the helicopter is. The genius of this though, is in the firmware. [Akshay] and [Sergio] are using an evolutionary algorithm adopted from Floreano et al, a noted author on biological inspired artificial intelligences. The idea is for the helicopter to create random “runs” and then check the data. The runs that are closer to the goal get refined while the others are eliminated, thus mimicking evolutions’ natural selection.

We’ve seen neural networks before, but nothing like this. Stay with us after the break, as we take this awesome project and narrow it down so that you too can implement this type of algorithm in your next project.

Continue reading “Self-Learning Helicopter Uses Neural Network” →

The Computer Without A CPU

April 20, 2014 by Brian Benchoff 31 Comments

1bit

[Jeff Laughton] was contacted by a customer that was interested in adding some automated functions to a printing press. Before eventually settling on a microcontroller for the job, [Jeff] went old school and started looking at logic gates, counters, and flip-flops. This lead him to the Motorola 14500 industrial control unit, a minimal processor with only 16 instructions. After a few ‘back of the napkin’ sketches, he came up with an extremely minimal computer that doesn’t use a microprocessor. It’s an interesting design notable not only for its electronic brevity, but also because it only uses one instruction.

The only instruction this computer will ever execute is an input test, the result of which controls a two-way branch. Instructions consist of an input address, output address, and a single bit of data. If the data bit is true, the computer jumps to one location in ROM, and if the data bit is false, a jump to another location is executed.

A computer really isn’t a computer without some form of memory, and this design is no exception. [Jeff] managed to add two bits of data between the 8-bit latch and 8-bit multiplexer in the design. This is enough to call a few subroutines which test the I/O-mapped memory to decide what the next instruction should be.

It’s a truly bizarre design, but actually much closer to a true Turing machine than the computers in your pocket, on your wrist, on your desk, and in your car.

Thanks [James] for the tip!

The Pyro Board: A Two Dimensional Ruben’s Tube

April 19, 2014 by James Hobson 36 Comments

Like visualizing music? Love fire? If so, you’re going to want to take a look at this Pyro Board.

What happens when you take a tube, put some holes along it, add a speaker on one end, pump some propane in, and then light it on fire? You get an awesome fire visual — also known as a Ruben’s Tube. It works because the sound pressure from the speakers causes the flow rate of gas leaving the holes to vary, which results in a visible “standing frequency” of flames, i.e. a flaming VU meter.

The folks over at [Fysikshow] decided to step it up a notch by building a 2-dimensional Ruben’s tube with 2500 holes. They have a steel box with the evenly spaced holes on the top, and two speakers attached to the sides. And it works amazingly well — see for yourself after the break.

Continue reading “The Pyro Board: A Two Dimensional Ruben’s Tube” →

Homemade Gravity Light Doesn’t Last Long But Proves The Concept!

April 16, 2014 by James Hobson 77 Comments

gravity light

After being inspired by the Deciwatt Gravity light, [Steve Dufresne] decided he wanted to try making his own as a proof of concept.

The Gravity Light by Deciwatt is an innovative device designed for third world countries to help eliminate expensive lighting like kerosene lamps. It has a small weight on a pulley which can be lifted up in under 3 seconds. During its slow descent down the weight provides light for 25 minutes! It’s affordable, sustainable, and reliable. It’s also mechanically impressive, which is exactly why [Steve] decided to try making his own.

He’s using a single LED, a small DC motor, a few pieces of wood, an old bicycle wheel, some bicycle chain, and a few jugs of water. The water is connected to the chain which is looped over the smallest gear on the bike. The generator is then powered by a belt wrapping around the outside of the rim. This gives the motor enough speed to generate electricity for the LED. His current design only lasts for about 3 minutes, but he’s already working on the second iteration. Testing systems like this really give you an appreciation for the effort that must have gone into the real Gravity Light.

Stick around after the break to see it in action.

Continue reading “Homemade Gravity Light Doesn’t Last Long But Proves The Concept!” →

Inkjet Transfers To Wood

April 16, 2014 by Adam Fabio 26 Comments

Color Image on wood board

You can’t feed a piece of wood through a stock inkjet printer, and if you could it’s likely the nature of the material would result in less than optimal prints. But [Steve Ramsey] has a tutorial on inkjet transfers to wood over on his YouTube Channel which is a simple two-step method that produces great results. We really love quick tips like this. Steve explains the entire technique while creating an example project – all in under 2 minutes of video. We don’t want to get your hopes up though – this method will only work on porous absorbent surfaces like bare wood, not on PC boards. We’ve featured some great Inject PCB resist methods here in the past though.

The transfer technique is dead simple. [Steve] uses the backing from a used sheet of inkjet labels (the shiny part that normally gets thrown away). He runs the backing sheet through his inkjet printer. Since plastic coated backing sheet isn’t porous, the ink doesn’t soak in and dry. He then presses the still wet page onto a piece of wood. The wet ink is instantly absorbed into the wood. A lacquer clear coat seals the image in and really make the colors pop. We’d like to see how this method would work with other porous materials, like fabrics (though the ink probably wouldn’t survive the washing machine).

Click past the break for another example of [Steve’s] work, and two videos featuring the technique.

Continue reading “Inkjet Transfers To Wood” →

MountainBeest – A Theo Jansen Creature Comes Alive In My Garage

April 14, 2014 by Jeremy Cook 17 Comments

About a year ago, a member of my family sent me a video featuring [Theo Jansen’s] StrandBeest, knowing that I was interested in all kinds of wacky and hackish inventions. My initial reaction was something to the effect of “wow that’s a neat device, but that guy is a little crazy.” For better or worse, the idea that this was an incredible invention turned over in my head for some time. Eventually, I decided that I needed to build one myself. Apparently I’m a little crazy as well.

Theo’s original beest runs on a complicated linkage system powered by wind. He was nice enough to publish the linkage lengths or “eleven holy numbers,” as he calls him at the bottom of this page. He doesn’t, however, really explain how the connections on his PVC power transmission system work, so I was left to try to figure it out from his videos. As you’ll see from build details and video to follow, this isn’t trivial. Keep reading past the jump to learn the adversity that I encountered, and how it was overcome in the end.

Continue reading “MountainBeest – A Theo Jansen Creature Comes Alive In My Garage” →

Electron Beam Control In A Scanning Electron Microscope

April 13, 2014 by Brian Benchoff 6 Comments

Electron

A few years ago [Ben Krasnow] built a scanning electron microscope from a few parts he had sitting around. He’s done a few overviews of how he built his SEM, but now he’s put up a great video on how to control electrons, focus them into a point, and scan a sample.

The basic idea behind a scanning electron microscope is to shoot electrons down a tube, focus them into a point, and scan a conductive sample and detect the secondary electrons shot off the sample and display them on an oscilloscope. [Ben] is generating electrons with a small tungsten filament at the top of his electron ‘stack’. Being like charged, these electrons naturally fan out, so a good bit of electron optics are required to get a small point.

Focusing is done through a series of pinholes and electrostatic deflectors, much like you’d see in an old oscilloscope CRT. In the video, you can see [Ben] shooting electrons and displaying a Christmas tree graphic onto a piece of phosphor-coated glass. He has a pretty big scanning area in his SEM, more than enough to look at a few chips, wafers, and whatever other crazy stuff is coming out of [Ben]’s lab.

Video below, along with the three-year-old overview of the entire microscope.

Continue reading “Electron Beam Control In A Scanning Electron Microscope” →