The seminal 1993 first-person shooter from id Software, DOOM, has become well-known as a test of small computer platforms. We’ve seen it on embedded systems far and wide, but we doubt we’ve ever seen it consume as little power as it does on a specialized neural network processor. The chip in question is a Syntiant NDP200, and it’s designed to be the always-on component listening for the wake word or other trigger in an AI-enabled IoT device.
DOOM running on as little as a milliwatt of power makes for an impressive PR stunt at a trade show, but perhaps more interesting is that the chip isn’t simply running the game, it’s also playing it. As a neural network processor it contains the required smarts to learn how to play the game, and in the simple circular level it’s soon picking off the targets with ease.
We’ve not seen any projects using these chips as yet, which is hardly surprising given their niche marketplace. It is however worth noting that there is a development board for the lower-range sibling chip NDP101, which sells for around $35 USD. Super-low-power AI is within reach.
If you’re making a lot of wiring harnesses, wrapping them can become a bit of a drag. [Well Done Tips] wanted to make this process easier and built a wiring harness wrapping machine.
The “C” shape of this wrapping machine means that you can wrap wires that are still attached at one or both ends, as you don’t have to pull the wires all the way through the machine. The plastic “C” rotates inside a series of pulleys with three of them driven by a belt attached to an electric motor. A foot pedal actuates the motor and speed is controlled by a rotary dial on the motor controller board.
Since this is battery powered, you could wrap wires virtually anywhere without needing to be near a wall outlet. This little machine seems like it would be really great if you need to wrap a ton of wire and shouldn’t be too complicated to build. Those are some of our favorite hacks.
It’s probably safe to say that most of us have had enough of the Great Balloon Follies to last the rest of 2023 and well beyond. It’s been a week or two since anything untoward was spotted over the US and subsequently blasted into shrapnel, at least that we know of, so we can probably put this whole thing behind us.
But as a parting gift, we present what has to be the best selfie of the year — a photo by the pilot of a U-2 spy plane of the balloon that started it all. Assuming no manipulation or trickery, the photo is remarkable; not only does it capture the U-2 pilot doing a high-altitude flyby of the balloon, but it shows the shadow cast by the spy plane on the surface of the balloon.
The photo also illustrates the enormity of this thing; someone with better math skills than us could probably figure out the exact size of the balloon from the apparent size of the U-2 shadow, in fact.
The end result is slightly reminiscent of embedding 3D printed shapes into tulle in order to create fantastic, armor-like flexible creations. But using rubber bands means the result is stretchy and compliant to a degree we haven’t previously seen. Keep it in mind the next time you’re trying to solve a tricky design problem; an embedded o-ring or rubber band might just do the trick.
Everyone knows we’re big fans of displays that differ from the plain old flat-panel LCDs that seem to adorn most devices these days. It’s a bit boring when the front panel of your widget is the same thing you stare at hour after hour while using your phone. Give us the chunky, blocky goodness of a vacuum fluorescent display (VFD) any day of the week for visual interest and retro appeal.
From the video below, it seems like [Posy] certainly is in the VFD fandom too, rolling out as he does example after example of unique and complicated displays, mostly from audio equipment that had its heyday in the 1990s. In some ways, the video is just a love letter to the VFD, and that’s just fine with us. But the teardowns do provide some insights into how VFDs work, as well as suggest ways to tweak the overall look of a VFD.
For example, consider the classy white VFDs that graced a lot of home audio gear back in the day. It turns out, the phosphors used in those displays weren’t white, but closer to the blue-green color that VFDs are often associated with. But put a pink filter between the display and the world, and suddenly those turquoise phosphors look white. [Posy] does a lot of fiddling with the stock filters to change the look of his VFDs, some to good effect, others less so.
As for the internals of VFDs, [Posy]’s look at a damaged display reveals a lot about how they work. With a loose scrap of conductor shorting one of the cathodes inside the tube, the damaged VFD isn’t much to look at, and is beyond reasonable repair, but it’s kind of cool to examine the spring mechanisms that take up slack as the cathodes heat up and expand.
Thanks to [Posy] for this heartfelt look into the VFDs of yesterday. If you need more about how VFDs work, we’ve covered that before, too.
If we were to talk to engineers about the childhood toys which most inspired them, it’s likely that the older among them would mention either Meccano or Erector Set. These similar construction toys using metal components originated independently around the turn of the 20th century in both Britain and America, and eventually became part of the same company.
It’s fair to say that the possibilities of those perforated metal sheets and myriad nuts and bolts might seem a little limited for the 2020s child, but it opens the age-old question of what remains to interest young minds in engineering or technology. The obvious answer to that question comes in the form of Lego, evidently so much more fun can be had with plastic bricks.
The basic concept is straightforward: enhance training data with hallucinated elements to change details, add variations, or introduce novel distractions. Studies show a robot additionally trained on this data performs tasks better than one without.
Suppose one has a dataset consisting of a robot arm picking up a coke can and placing it into an orange lunchbox. That training data is used to teach the arm how to do the task. But in the real world, maybe there is distracting clutter on the countertop. Or, the lunchbox in the training data was empty, but the one on the counter right now already has a sandwich inside it. The further a real-world task differs from the training dataset, the less capable and accurate the robot becomes.
ROSIE aims to alleviate this problem by using image diffusion models (such as Imagen) to enhance the training data in targeted and direct ways. In one example, a robot has been trained to deposit an object into a drawer. ROSIE augments this training by inpainting the drawer in the training data, replacing it with a metal sink. A robot trained on both datasets competently performs the task of placing an object into a metal sink, despite the fact that a sink never actually appears in the original training data, nor has the robot ever seen this particular real-world sink. A robot without the benefit of ROSIE fails the task.