Neural networks ought to be very appealing to hackers. You can easily implement them in hardware or software and relatively simple networks can perform powerful functions. As the jobs we ask of neural networks get more complex, the networks require more artificial neurons. That’s why researchers are pursuing dense integrated neuron chips that could do for neural networks what integrated circuits did for conventional computers.
Researchers at Princeton have announced the first photonic neural network. We recently talked about how artificial neurons work in conventional hardware and software. The artificial neurons look for inputs to reach a threshold which causes them to “fire” and trigger inputs to other neurons.
To map this function to an optical device, the researchers created tiny circular waveguides in a silicon substrate. Light circulates in the waveguide and, when released, modulates the output of a laser. Each waveguide works with a specific wavelength of light. This allows multiple “inputs” (in the form of different wavelengths) to sum together to modulate the laser.
The team used a 49-node network to model a differential equation. The photonic system was nearly 2,000 times faster than other techniques. You can read the actual paper online if you are interested in more details.
There’s been a lot of work done lately on both neural networks and optical computing. Perhaps this fusion will advance both arts.
At its heart is an Arduino Uno and an Adafruit Motor Shield v2. The green laser is turned on and off by the Arduino through a transistor. But the part that makes this really a fun machine to watch at work are the two stepper motors and two mirrors that reflect the laser in the X and Y directions. The mirrors are rectangles cut from a hard disk platter, which if you’ve ever seen one, is very reflective. The servos tilt the mirrors at high speed, fast enough to make the resulting projection on the wall appear almost a solid shape, depending on the image.
He’s even written a Windows application (in C#) for remotely controlling the projector through bluetooth. From its interface you can select from around sixteen predefined shapes, including a what looks like a cat head, a heart, a person and various geometric objects and line configurations.
There is a sort of curving of the lines wherever the image consists of two lines forming an angle, as if the steppers are having trouble with momentum, but that’s probably to be expected given that they’re steppers controlling relatively large mirrors. Or maybe it’s due to twist in the connection between motor shaft and mirror? Check out the video after the break and let us know what you think.
There’s just something about wielding a laser pointer on a dark, foggy night. Watching the beam cut through the mist is fun – makes you feel a little Jedi-esque. If you can’t get enough of lasers and mist, you might want to check out this DIY “laser sky” effect projector.
The laser sky effect will probably remind you of other sci-fi movies – think of the “egg scene” from Alien. The effect is achieved by sweeping a laser beam in a plane through swirling smoke or mist. The laser highlights a cross section of the otherwise hidden air currents and makes for some trippy displays. The working principle of [Chris Guichet]’s projector is simplicity itself – an octagonal mirror spun by an old brushless fan motor and a laser pointer. But after a quick proof of concept build, he added the extras that took this from prototype to product. The little laser pointer was replaced with a 200mW laser module, the hexagonal mirror mount and case were 3D printed, and the mirrors were painstakingly aligned so the laser sweeps out a plane. An Arduino was added to control the motor and provide safety interlocks to make sure the laser fires only when the mirror is up to speed. The effect of the deep ruby red laser cutting through smoke is mesmerizing.
Most data storage devices we currently use are, at their core, two-dimensional. Sure, a hard drive might have multiple platters, but the data storage takes place on a flat surface. Even an optical drive is effectively a single surface that holds data. At the City College of New York, they are experimenting with storing data in three dimensions using lab-grown diamonds and LASERs.
Usually, diamonds that have few flaws are more valuable. But in this application, the researchers exploit the flaws to store information. Optical memory that uses a volume instead of a surface isn’t exactly new. However, it is difficult to use these techniques in a way that is rewritable.
Diamonds are a crystalline structure of carbon atoms. Sometimes, though, a carbon atom is missing from the structure. That’s a vacancy. Another defect is when a nitrogen atom replaces a carbon atom. Sometimes a vacancy occurs next to a rogue nitrogen atom and that causes an NV (nitrogen vacancy) center.
Everyone knows you can’t visibly bend light over short distances in free air. Or can you? [Jack Pearse] has figured out a way to do it though, or at least make it appear that way. He does it by combining a trick of math and a trick of the eye. The secret is the hyperboloid, a geometric construct described by a quadratic equation. [Jack’s] creation is more specifically a hyperboloid in one sheet. This type of structure allows straight lines to create a an overall curved surface. Hyperboloids have been used by architects and in construction for years, often in tall structures like water towers.
If a bunch of straight steel beams can form a curved shape, lasers should be able to pull off the same effect. By employing persistence of vision, [Jack] was able to create his hyperboloid with only 10 small lasers. The lasers are mounted on the rim of a bicycle wheel and carefully aimed. The wheel is spun up with using an electric bicycle motor. [Jack] kept things safe by building a centrifugal switch. The switch powers up all the lasers in when the tire is spinning. This ensures no one can be hit by a static beam.
Once the wheel is spinning, all you need is a bit of smoke or haze in the room. The spinning lasers combine to form the hyperboloid shape. You can see the project in action in the video after the break.
[Mathieu] wrote in with his laser target practice game. It’s not the most amazing hack in the history of hackery, but it’s an excellent example of the type of simple and fun things you can do with just a little bit of microcontrollering.
First off, the gun is a broken toy gun that used to shoot something other than red collimated light beams. The Arduino knockoff inside reacts to a trigger pull and fires the laser for around 200 milliseconds. The gun also has a “gas gauge” that fills up with repeated shots and cools down over time. And therein lies the game — a simple race to ten, where each player only has a fixed number of shots over time.
The targets are simply a light sensor, scorekeeping LED display, and a buzzer that builds tension by beeping at you as the countdown timer ticks down. The bodies are made out of 3D-printed corners that connect some of [Mathieu]’s excess wooden goat-cheese lids.
All the code is up on GitHub so you can make your own with stuff that you’ve got lying around the house. The “gun” can be anything that you can embed a laser in that makes it aimable. Good clean fun!
Well all know cellular automata from Conway’s Game of Life which simulates cellular evolution using rules based on the state of all eight adjacent cells. [Gavin] has been having fun playing with elementary cellular automata in his spare time. Unlike Conway’s Game, elementary automata uses just the left and right neighbors of a cell to determine the next cell ahead in the row. Despite this comparative simplicity, some really complex patterns emerge, including a Turing-complete one.
[Gavin] started off doing the calculations by hand for fun. He made some nice worksheets for this. As we can easily imagine, doing the calculations by hand got boring fast. It wasn’t long before his thoughts turned to automating his cellular automata. So, he put together an automatic cellular automator. (We admit, we are having a bit of fun with this.)