Disney’s New Robot Limbs Trained Using Neural Networks

July 3, 2018 by Steven Dufresne 9 Comments

Disney is working on modular, intelligent robot limbs that snap into place with magnets. The intelligence comes from a reasonable sized neural network that also incorporates some modularity. The robot is their Snapbot whose base unit can fit up to eight of limbs, and so far they’ve trained with up to three together.

The modularity further extends to a choice of three types of limb. One with roll and pitch, another with yaw and pitch, and a third with roll, yaw, and pitch. Interestingly, of the three types, the yaw-pitch one seems most effective.

In this age of massive, deep neural networks requiring GPUs or even online services for training in a reasonable amount of time, it’s refreshing to see that this one’s only two layers deep and can be trained in three hours on a single-core, 3.4 GHz Intel i7 processor. Three hours may still seem long, but remember, this isn’t a simulation in a silicon virtual world. This is real-life where the servo motors have to actually move. Of course, they didn’t want to sit around and reset it after each attempt to move across the table so they built in an automatic mechanism to pull the robot back to the starting position before trying to cross the table again. To further speed training, they found that once they’d trained for one limb, they could then copy the last of the network’s layers to get a head starting on the training for two limbs.

Why do training? Afterall, we’ve seen pretty awesome multi-limbed robots working with manual coding, an example being this hexapod tank based on one from the movie Ghost in the Shell. They did that too and then compared the results of the manual approach with those of the trained one and the trained one moved further in the same amount of time. At a minimum, we can learn a trick or two from this modular crawler.

Check out their article for the details and watch it in action in its learning environment below.

Continue reading “Disney’s New Robot Limbs Trained Using Neural Networks” →

Roboshield Helps Your Robot Walk And Talk

July 3, 2018 by Lewin Day 3 Comments

The joy of building robots comes from being able to imbue them with as much or as little personality and functionality as you wish during the design and build process. While creative flair and originality is always a good thing, there’s a lot of basic needs many robots have in common with each other, so where possible it’s good to avoid reinventing the wheel so more time can be spent on more advanced features. Roboshield aims to help make the basics easy so you can let your robot freak flag fly!

At its core, it’s an Arduino shield that packs in a host of hardware to get your robot up and running. As far as motion is concerned, a PCA9685 module is used to allow the control of 8 servos, plus there’s a TB6621FNG dual motor speed controller that offers both speed control and forward/reverse. That’s enough to get your electronic buddy scooting about the floor and waving its arms in the air.

The party piece, however, is the Vstamp text-to-speech module. This device produces a beautiful cliche electronic voice, which your robot is legally required to use to recite Asimov’s Laws of Robotics. Overall, it’s a tidy project that can take the hassle out of getting your robot design up and running, leaving you to focus on the fun bits like death rays and tractor beams. We can’t wait to see it powering the next wave of sassy DIY robots.

Philo Farnsworth, RCA, And The Battle For Television

July 3, 2018 by Dan Maloney 35 Comments

The parenthood of any invention of consequence is almost never cut and dried. The natural tendency to want a simple story that’s easy to tell — Edison invented the light bulb, Bell invented the telephone — often belies the more complex tale: that most inventions have uncertain origins, and their back stories are often far more interesting as a result.

Inventing is a rough business. It is said that a patent is just a license to get sued, and it’s true that the determination of priority of invention often falls to the courts. Such battles often pit the little guy against a corporate behemoth, the latter with buckets of money to spend in making the former’s life miserable for months or years. The odds are rarely in the favor of the little guy, but in few cases was the deck so stacked against someone as it was for a young man barely out of high school, Philo Farnsworth, when he went up against one of the largest companies in the United States to settle a simple but critical question: who invented television?

Continue reading “Philo Farnsworth, RCA, And The Battle For Television” →

A Cartoon-ifying Camera For Instant Absurdism

July 3, 2018 by Kristina Panos 12 Comments

We take photographs as a way to freeze moments in time and to capture the details that get blurred by our unreliable memories. There is little room for interpretation, and this is kind of the whole point.

[Dan Macnish]’s latest project, Draw This, turns reality into absurdity. It’s a Raspberry Pi-based instant camera that trades whatever passed in front of the lens for a cartoon version of same. Draw This uses neural networks to ID the objects in the frame, and then draws upon thousands of images from Google’s Quick, Draw! dataset to provide a loose interpretation via thermal printer. Seems to us like the perfect camera to take to DEFCON (or any other part of Las Vegas).

If you have a Raspi3, a v2 camera, and a thermal printer, you can make your own crowd-sourced, cartoonified memories using the code in [Dan]’s repo. Still into recording reality? You can use Pi cameras to see in the dark or even explore a body of water.

These Capacitors Are A Cheap Gimmick

July 3, 2018 by Al Williams 31 Comments

If you search through an electrical engineering textbook, you probably aren’t going to find the phrase “gimmick capacitor” but every old ham radio operator knows about them. They come in handy when you need a very small capacitor of unknown value. For example, if you are trying to balance the stray capacitance in a circuit, you might not know exactly what value you need, but you know it won’t be very much. That’s when you want a gimmick capacitor.

A gimmick capacitor is made by taking two strands of insulated wire and twisting them together; the length and the tightness of the twist determine the capacitance. Tightening or loosening the twist, or trimming some of the wire off, makes it tunable.

These are most commonly found in RF equipment or high-speed logic because of the small capacitance involved — usually about 1 to 2 pF per inch of twist or so. The thicker the insulation, the less capacitance you’ll get, so it is common to use magnet wire or something else with a thin insulating layer. You can take this one step further and decrease the spacing by stripping down one wire as long as it isn’t going to touch anything else.

Obviously, the insulation needs to be good enough for the voltage on them, an important consideration in tube circuits, for instance. But other than that, a gimmick capacitor is a straightforward tool to have in your box of design tricks. Can we take this further? Continue reading “These Capacitors Are A Cheap Gimmick” →

Drawing Lines In The Sand: Taking Beach Graffiti To The Next Level

July 3, 2018 by Ben James 39 Comments

When strolling down the beach, there’s always an urge to draw in the sand – it seems compulsory to make your mark by inscribing something. But there’s a dilemma: how do you go about physically drawing it? You could opt to remain standing and attempt to deploy a toe, but that requires a level of dexterity few possess. The only other option is to bend down and physically use your hands. Ultimately, there’s no way to draw anything in the sand without losing your dignity.

The solution? A robot, of course – the brainchild of [Ivan Miranda]. The idea is simple and elegantly executed: make a large linear actuator, place it on wheels, and attach a servo which can position an etching tool to be either in the sand or above it. The whole contraption moves forward one column at a time, making a vertical pass with the marker being engaged or disengaged as required. The columns are quite thin, giving relatively high-resolution text, though this does mean it take a while. Adding another servo and marking two adjacent columns at the same time would be an easy way to instantly double the speed.

The wheels are big and chunky, to ensure the horizontal distance travelled does not change between the top and the bottom. Of course, when making big parts like these it always helps if you’ve already built a giant custom 3D printer. If you want to read more of [Ivan]’s large scale 3D printing antics, checkout his tank with suspension, or plus-sized seven-segment clock.

Continue reading “Drawing Lines In The Sand: Taking Beach Graffiti To The Next Level” →

The A To Z Of Building Your Own Keyboard

July 3, 2018 by Tom Nardi 6 Comments

We’ve featured a number of people who’ve taken the plunge and created their own customized keyboard; at this point it’s safe to say that there’s enough information and source code out there that anyone who’s looking to build their own board won’t have much trouble figuring out how to do so. That being said, it’s nice to have a comprehensive at a process from start to finish. Why sift through forum posts and image galleries looking for crumbs if you don’t have to?

That’s precisely what makes this write-up by [Maarten Tromp] so interesting. He walks the reader through every step of the design and creation of his customized keyboard, from coming up with the rather unique layout to writing the firmware for its AVR microcontroller. It’s a long read, filled with plenty of tips and tricks from a multitude of disciplines.

After looking at other custom boards for inspiration, [Maarten] used OpenSCAD to create a 3D model of his proposed design, and had it printed at Shapeways. His electronics are based around an Atmel ATMega328P using vUSB, and Microchip MCP23017 I/O expanders to connect all the keys. He wrapped it all up by designing a PCB in gEDA PCB and having it sent off for production. As a testament to his attention to detail, everything mated up on the first try.

[Maarten] is happy with the final product, but mentions that in a future revision he would like to add RGB lighting and use a microcontroller that has native USB support. He’d also like to drop the I/O expanders and switch over to Charlieplexing for the key matrix.

From uncommon layouts to diminutive technicolor beauties, it seems there’s no end of custom keyboards in sight. We aren’t complaining.