Measuring Trees Via Satellite Actually Takes A Great Deal Of Field Work

Figuring out what the Earth’s climate is going to do at any given point is a difficult task. To know how it will react to given events, you need to know what you’re working with. This requires an accurate model of everything from ocean currents to atmospheric heat absorption and the chemical and literal behavior of everything from cattle to humans to trees.

In the latter regard, scientists need to know how many trees we have to properly model the climate. This is key, as trees play a major role in the carbon cycle by turning carbon dioxide into oxygen plus wood. But how do you count trees at a continental scale? You’ll probably want to get yourself a nice satellite to do the job.

Continue reading “Measuring Trees Via Satellite Actually Takes A Great Deal Of Field Work”

A Straightforward AI Voice Assistant, On A Pi

With AI being all the rage at the moment it’s been somewhat annoying that using a large language model (LLM) without significant amounts of computing power meant surrendering to an online service run by a large company. But as happens with every technological innovation the state of the art has moved on, now to such an extent that a computer as small as a Raspberry Pi can join the fun. [Nick Bild] has one running on a Pi 4, and he’s gone further than just a chatbot by making into a voice assistant.

The brains of the operation is a Tinyllama LLM, packaged as a llamafile, which is to say an executable that provides about as easy a one-step access to a local LLM as it’s currently possible to get. The whisper voice recognition sytem provides a text transcript of the input prompt, while the eSpeak speech synthesizer creates a voice output for the result. There’s a brief demo video we’ve placed below the break, which shows it working, albeit slowly.

Perhaps the most important part of this project is that it’s easy to install and he’s provided full instructions in a GitHub repository. We know that the quality and speed of these models on commodity single board computers will only increase with time, so we’d rate this as an important step towards really good and cheap local LLMs. It may however be a while before it can help you make breakfast.

Continue reading “A Straightforward AI Voice Assistant, On A Pi”

Gold Recovery From E-Waste With Food-Waste Amyloid Aerogels

A big part of the recycling of electronic equipment is the recovery of metals such as gold. Usually the printed circuit boards and other components are shredded, sorted, and then separated. But efficiently filtering out specific metals remains tricky and adds to the cost of recycling. A possible way to optimize the recovery of precious metals like gold could be through the use of aerogels composed out of protein amyloids to which one type of metal would preferentially adsorb. According to a recent research article in Advanced Materials by [Mohammad Peydayesh] and colleagues, such aerogels could be created from protein waste from the food industry.

The adsorption mechanism of the protein amyloids is a feature of these proteins which form chelants, which are structures that can effectively bond to metal ions. These are usually organic compounds, and are used in certain medical treatments where heavy metal poisoning is involved (chelation therapy). By having these protein amyloids in an aerogel structure, the surface area for adsorption is maximized, which in the research article is said to have an efficiency of 93.3% for gold recovery, while leaving the other metals in the aqua regia solution (nitric and hydrochloric acid) mostly untouched.

Of note here is that although the food waste protein angle is taken, the experiment used whey protein. This is also one of the most popular food supplements in the world, to the point that microbial production of whey is a thing now. Although this doesn’t invalidate the aerogel chelation approach to e-waste recycling, it’s a curious omission in the article that does not appear to be addressed.

Making The Commodore SX-64 Mini

When you find a portable TV from the 1980s, and it reminds you of the portable Commodore 64, there’s only one thing to be done. [Aaron Newcomb] brings us the story of taking an Emerson PC-6 and mating it to the guts of his THEC64 Mini. It’s a bit of a journey, as the process includes modding the TV to include a composite input and trimming some unused PCB off the TV’s mainboard. Then some USB ports and a three-and-a-half inch floppy drive were shoehorned into the chassis, with the rear battery compartment holding the parts from THEC64 Mini.

The build was not entirely without issue. It turns out the degaussing coil connector can plug perfectly into the service port, and Murphy’s law proved itself true again. But no harm was done, and the error was quickly discovered. All that was left was to button the chassis back up and add some paint and 3d-printed trim details. The build looks great! Come back after the break to watch the video from the [Retro Hack Shack] for yourself.

Continue reading “Making The Commodore SX-64 Mini”

ForceGen: Using A Diffusion Model To Help Design Novel Proteins

Although proteins are composed out of only a small number of distinct amino acids, this deceptive simplicity quickly vanishes when considering the many possible sequences across a protein, not to mention the many ways in which a single 1D protein sequence can fold into a 3D protein shape with a specific functionality. Although natural evolution has done much of the legwork here already, figuring out new sequences and their functionality is a daunting task where increasingly deep learning algorithms are being applied. As [Bo Ni] and colleagues report in a research article in Science Advances, the hardest challenge is designing a protein sequence based on the desired functionality. They then demonstrate a way to use a generative model to speed up this process.

They set out to design proteins with specific mechanical properties, for which they used the known unfolding characteristics of various protein sequences to train a diffusion model. This approach is thus more akin to the technology behind image generation algorithms like DALL-E than LLMs. Using the trained diffusion model it was then possible to generate likely sequences of which the properties could then be simulated, with favorable results.

As a large data set aid, such a diffusion model could conceivably be very useful in fields even beyond protein synthesis, automating tedious tasks and conceivably speeding up discoveries.

Watch The OpenScan DIY 3D Scanner In Action

[TeachingTech] has a video covering the OpenScan Mini that does a great job of showing the workflow, hardware, and processing method for turning small objects into high-quality 3D models. If you’re at all interested but unsure where or how to start, the video makes an excellent guide.

We’ve covered the OpenScan project in the past, and the project has progressed quite a bit since then. [TeachingTech] demonstrates scanning a number of small and intricate objects, including a key, to create 3D models with excellent dimensional accuracy.

[Thomas Megel]’s OpenScan project is a DIY project that, at its heart, is an automated camera rig that takes a series of highly-controlled photographs. Those photographs are then used in a process called photogrammetry to generate a 3D model from the source images. Since the quality of the source images is absolutely critical to getting good results, the OpenScan hardware platform plays a pivotal role.

Once one has good quality images, the photogrammetry process itself can be done in any number of ways. One can feed images from OpenScan into a program like Meshroom, or one may choose to use the optional cloud service that OpenScan offers (originally created as an internal tool, it is made available as a convenient processing option.)

It’s really nice to have a video showing how the whole workflow works, and highlighting the quality of the results as well as contrasting them with other 3D scanning methods. We’ve previously talked about 3D scanning and what it does (and doesn’t) do well, and the results from the OpenScan Mini are fantastic. It might be limited to small objects, but it does a wonderful job on them. See it all for yourself in the video below.

Continue reading “Watch The OpenScan DIY 3D Scanner In Action”

Installing SteamOS And Windows On A Google Meet Video Conference Computer

The Lenovo Meet is a collaboration with Google to bring Google Meet to customers in a ready to install kit for conference rooms and similar. Also called the Google Meet Series One, it features a number of cameras, speakers, display and more, along with the base unit. It is this base unit that [Bringus Studios] on YouTube tried to install a different OS capable of running Steam games on in a recent video. Along the way many things were learned about this device, which is – unsurprisingly – just another ChromeOS box.

After removing the rubber bottom (which should have been softened with a hot air gun to prevent damage), the case can be opened with some gentle prying to reveal the laptop-like innards. Inside are an 8th gen Intel CPU (i7-8550U @ 1.8 GHz), a 128 GB SATA M.2, 2 GB DDR4 RAM, along with 2 more GB of DDR4 a MicroSD slot and a Google Coral DA1 TPU on the bottom of the mainboard. It should be easy to install Linux, Windows, etc. on this other than for the ChromeOS part, which locks down the non-UEFI BIOS firmware.

Continue reading “Installing SteamOS And Windows On A Google Meet Video Conference Computer”