Auto Added by WPeMatico
GitHub user jomjol posts a project for performing Artificial Intelligence (AI) calculations on a very cheap hardware (less than ten euros).
It uses an ESP32 with a camera interface, so it’s very small (4.4 x 2.9 x 2cm).
All the code is located on GitHub under MIT license and the 3D case is on Thingiverse.
See details in the wiki here.
Peter Naftaliev is an artificial intelligence (AI) researcher and consultant who is working in making 2D into 3D. Besides work for his company Abelians, he also publishes breakthrough research at his blog here and uses his 15 years of machine learning experience to teach the technology. If you’re interested in Project Management for Machine Learning, or were looking for a Deep Learning Dictionary, or want to write your own patent, Naftaliev is a wealth of information.
Naftaliev first came to our attention when he published work showing us how to take a 2D image and turn it into a possibly 3D-printed file. Creating 3D content, moreover 3D printable 3D content is difficult. CAD is still too complex for most people and 3D scanning works, but is finicky. Everyone can either draw or take photos that lead to digital 2D content, however. If we could easily take 2D data and make it 3D printable, one could much more easily make their own 3D-printed products, let consumers mass-customize things or make custom-fit things like shoe soles.
So, Naftaliev’s work on the cutting edge of AI and 3D printing is, to me, of potential crucial importance to the future of 3D printing. Eventually raw computing power, improving cameras and better software could help us get to a stage where all of our phones are 3D scanners that can be used to create 3D content easily. Until then, and also subsequently, AI and machine learning could let us take much more content and make it 3D.
Machine learning and AI, however, are kind of like a magical sauce that is supposed to make everything better for everyone all the time. I remember when 3D printing was seen in the same light. I personally tried to be a realistic, enlightening, but not dazzlingly optimistic guide for people through these hype times. For AI and the intersection of machine learning and 3D printing, Naftaliev is this person for me. Not for me alone, however.
Corona-bored Naftaliev posted on Reddit. He decided to do a Zoom call about deep learning. He had a previously prepared lecture which he couldn’t give because all of his lectures and conferences where he would speak were Corona-cancelled. So, he made a Reddit post about his online lecture. Just based on that one Reddit post, Naftaliev had 350 people sign up for his lecture. Now, he has a Reddit community that is all about “2D, 3D and AI. Video, image, 3D modeling, depth maps, neural networks.” You can subscribe to the newsletter here or the meetup here to be kept abreast of goings-on.
Enthused by this, he continued with talks on:
“image processing, AI, 3d modeling, technological advancements . But, more importantly, it is a step to try to democratize access to information to anyone around the world. Academic research and papers can be very hard to figure out even for people who are working in the subject. Reading just one paper and truly understanding what is going on can take several good days of work and sometimes requires access to people who have knowledge in the field. And, what’s more, a lot of the research does not come with an open source where you can try to test things out yourself (it is extremely hard to replicate the code and results of a paper, if not impossible because of access to training data and computational resources). The research that does come with code many times is still hard to figure out, sometimes there are bugs or things in the code that do not align with the research. I want to help make all of this more accessible to people everywhere.”
He feels that “if I or anyone else has put the time and effort to understand some new research that is out, why not share it with others.” He does each live lecture twice, once for the east side of the planet, once for the west. He then offers these lectures for download. The next lecture deals with generating art using neural networks.
In the future he hopes “to get the authors of the most important researches in our field to come and present their papers, code and the latest advancements – live, online, for anyone who is interested in learning more.” The lectures are clear and super interesting but not necessarily for casual viewing, so paying attention helps. Naftaliev means for them to be for,
“People anywhere in the world with technical orientation who are interested in machine learning, or are already full-fledged practitionersresearchers who want to expand their understanding of sub-topics in this sphere. We are also touching the boundary of digital art, so people from the digital arts that want to see what the latest technological research in the field can do and how they can use it for their art.”
In terms of background,
“Mathematical and programming background is a plus. We do explain basic concepts in machine learning if we see that the audience is not fully familiar with them. Every participant who signs in to listen to a lecture fills out a small bio about himself so we know how much intro material we need to explain and how deep we can dive.
Naftaliev hopes that you can learn,
“Which papers and open source findings are interesting and relevant, the current state of the art results and how to replicate them, current technological limitations in industry and academia, expend your horizons about what is possible to achieve with AI in everything to do with image processing, 3D modeling, signal processing and more. I am also experimenting with allowing people to get to know each other and form connections around the world by sharing these common interests.”
You can find the YouTube channel here. Below you can see how you can go from 2D to 3D using neural nets.
I think that this is fascinating and, with Naftaliev’s help, you can be transported to the cutting edge of making 2D 3D and understand more about machine learning. I really think that this is an emerging frontier for our industry and am very grateful that Naftaliev will be giving this series of lectures. Subscribe here.
A lecture by a guest Dr. Eyal Gruss, Fake Anything the Art of Deep Learning is here.
The post Peter Naftaliev Lecture Series on Learning on How to Go from 2D to 3D with Machine Learning appeared first on 3DPrint.com | The Voice of 3D Printing / Additive Manufacturing.
Researchers from Tsinghua University are investigating artificial intelligence and all that surrounds it, from materials to workflow, performance, and all the elements of intelligent construction. Their experiments are outlined in the recently published, ‘Re-perceive 3D printing with Artificial Intelligence.’
Today, artificial intelligence (AI) is becoming even more of a reality, and its uses continue to travel far beyond imagery, speech, and the ability to make choices. Architects are expanding construction methods with AI, and within digital construction, positively disrupting conventional practices—and offering enormous potential. In this study, the researchers explore the possibilities of finding new, universal methods for AI—along with mapping material/distribution properties to material behavior.
Their new system can print a spatial wireframe either in 2.5 or 3D, and without supports. Because of intermolecular forces and more, it is not possible to achieve comprehensive control of the printing process with basic software. The researchers studied automatic systems and image processing in order to create a model calculating the printed code, while the other predicts G-Code.
Two-way mapping between Gcode and the printed form
The main workflow includes:
In model training, the researchers included:
Image to data
The ultimate goal is to propel a new manufacturing process forward, centered around the use of dynamic material properties. They report that their new model has ‘already achieved a positive result,’ with shapes printed accurately; however, there is still a concern as to whether they can predict a G-Code set for curves required for printing. Their new method expands the application of a scientific workflow and creates a direct end-to-end connection—from fabrication to the final form.
“… compared to the traditional method of developing material models that require different fields of knowledge and workflow for different material properties, the method and logic of the work described in this paper is universal and proven to be capable of generalization when applied to producing a diversity of material performance models that encompass multiple systems of non-related material behavior, such as the active process of bending elastomer and the reform process of melting thermoplastics.”
”For architects, architectural design ultimately relies on the selection and construction of materials. In the long run, new building materials and new construction methods will bring about tremendous changes in the construction industry, as well as new architectural styles. We believe that the approach presented in this article will be a positive inspiration to herald this change,” concluded the researchers.
Artificial intelligence is gaining further ground and accompanies 3D printing and other related technologies too as researchers around the world endeavor to seek further imaging recognition capabilities, use it as a measuring device in 3D printing, and even analyze cardiac health.
What do you think of this news? Let us know your thoughts! Join the discussion of this and other 3D printing topics at 3DPrintBoard.com.
Structure of
Convolutional Neural Networks (CNN)
[Source / Images: ‘Re-perceive 3D printing with Artificial Intelligence’]
The post Beijing: Researchers Improving 3D Printing with Artificial Intelligence appeared first on 3DPrint.com | The Voice of 3D Printing / Additive Manufacturing.
USEED is a Korean startup headed by Jung Soo Lim. The eight-person company got its start making 3D printer kits, and specializes in the education market. The firm makes robotics kits, Prusa i3 type 3D printers, its own Creator 3D printer, and even an SMT placement machine. The company designs, develops and manufactures its machines in Korea and has been expanding steadily. Now they aim to undertake a bit of a quantum leap. The firm has been designing and testing its Thingi for months now. The Thingi is one of the, if not the most, adorable 3D printers I’ve ever seen. Specially designed to be accessible, safe and easy to use for children, the Thingi is meant to let kids easily 3D print. The 125 x 140 x 190 build volume machine can print over WiFi, has a 260C capable extruder, and can print up to 60 mm/s. The printer has a new trick up its sleeve, however. The voice-activated AI-powered printer can listen to kids’ commands and prints accordingly. If the voice activation works well and the company can accordingly automate the entire printing workflow, it would make 3D printing much more accessible and easier. Potentially it would make 3D printing much easier for all of us as well. It was refreshing to finally see something innovative happening again in desktop 3D printing. The company is testing the printer now and aims to go to crowdfunding in a few months. We interviewed CEO Jung Soo Lim to find out more.
What is USEED?
Our company manufactures 3D printers and coding education kits for kids. We have 8 employees. Our company was founded to provide IT seeds that can be easily implemented if anyone has an idea.
Therefore, we are currently supplying educational 3D printers and related education services to Korean educational institutions. The big plan that our company has is to launch a voice driven AI printer. In Korea, 3D printing is not being popularized fast enough. We analyzed the of this causes through our experiences in selling technical products. There was pressure to learn 3D modeling in order to use 3D printers. Also, children between the ages of 5 and 10 want to use 3D printers. However, these purchases were not made because children had to use computers independently. With that background, our company developed the Thingi early last year and is currently preparing to mass-produce it. Also, to cover the costs of molds, we are looking for funds through crowdfunding platforms.
Our company will complete 3D printers and content that children can use for making their own toys. We hope to have is available this year at Christmas.
Why a voice-activated 3D printer?
I want kids to have fun while using 3D printers to make the things that they want. If a child wants to make a Hello Kitty patterned cup, the kid talks to our 3Dprinter. The Thingi recommends the best model file and will then print it out. We will simplify this process so that children can use 3D printers in a fun way. I think that if many people enjoy and use our products, the limitations of technology will be overcome.
The post Korean Startup USEED to Launch Voice Activated AI Driven 3D Printer for Kids appeared first on 3DPrint.com | The Voice of 3D Printing / Additive Manufacturing.
We’ve got so much happening here at Adafruit that it’s not always easy to keep up! Don’t fret, we’ve got you covered. Each week we’ll be posting a handy round-up of what we’ve been up to, ranging from learn guides to blog articles, videos, and more.
Tiny Machine Learning on the Edge with TensorFlow Lite Running on SAMD51
Tiny Machine Learning on the Edge with TensorFlow Lite Running on SAMD51 (video and/or skip to the demo part at 7 min 28 secs). You’ve heard of machine learning (ML), but what is it? And do you have to buy specialty hardware to experiment? If you have some Adafruit hardware, you can build some Tiny ML projects today!
More BLOG:
Keeping with tradition, we covered quite a bit this past week. Here’s a kinda short nearing medium length list of highlights:
1,900th GUIDE! Trash Panda 2: Garbage Day
Our 1,900th guide has landed in the Learn System! It’s John Park’s Trash Panda 2: Garbage Day and it’s lots of fun! You can make it with MakeCode Arcade, and mod and hack it all you like.
In Trash Panda 2: Garbage Day, you play as the suburban dweller just trying to get some sleep when the raccoons and cats decide its time to make noise and throw garbage our of the trash bins! You must try to stop them by shining your flashlight on them. But you can only play at night, so be sure that your PyGamer or PyBadge’s light sensor indicates it’s dark out!
More LEARN:
Browse all that’s new in the Adafruit Learning System here!
From Jim Ewing, Lucas E on Hackster.io:
We took on the challenge of re-tasking a $5 Raspberry Pi Zero W computer-on-a-stick and a $25 Pi Camera using the latest software platforms available to create an intelligent security system that recognizes people in front of the camera and notifies the owner via SMS text messages or email.
Each Friday is PiDay here at Adafruit! Be sure to check out our posts, tutorials and new Raspberry Pi related products. Adafruit has the largest and best selection of Raspberry Pi accessories and all the code & tutorials to get you up and running in no time!
Artificial intelligence (AI) company Intellegens, which is a spin-off from the University of Cambridge, created a unique toolset that can train deep neural networks from noisy or sparse data. The machine learning algorithm, called Alchemite, was created at the university’s Cavendish Laboratory, and is now making it faster, easier, and less expensive to design new materials for 3D printing projects. The Alchemite engine is the company’s first commercial product, and was recently used by a research collaboration to design a new nickel-based alloy for direct laser deposition.
Researchers at the university’s Stone Group, along with several commercial partners, saved about $10 million and 15 years in research and development by using the Alchemite engine to analyze information about existing materials and find a new combustor alloy that could be used to 3D print jet engine components that satisfy the aerospace industry’s exacting performance targets.
“Worldwide there are millions of materials available commercially that are characterised by hundreds of different properties. Using traditional techniques to explore the information we know about these materials, to come up with new substances, substrates and systems, is a painstaking process that can take months if not years,” Gareth Conduit, the Chief Technology Officer at Intellegens, explained. “Learning the underlying correlations in existing materials data, to estimate missing properties, the Alchemite engine can quickly, efficiently and accurately propose new materials with target properties – speeding up the development process. The potential for this technology in the field of direct laser deposition and across the wider materials sector is huge – particularly in fields such as 3D printing, where new materials are needed to work with completely different production processes.”
Alchemite engine
Alchemite is based on deep learning algorithms which are able to see correlations between all available parameters in corrupt, fragmented, noisy, and unstructured datasets. The engine then unravels these data problems and creates accurate models that are able to find errors, optimize target properties, and predict missing values. Alchemite has been used in many applications, including drug discovery, patient analytics, predictive maintenance, and advanced materials.
Thin films of oxides deposited with atomic layer precision using pulsed laser deposition. [Image: Adam A. Læssøe]
“Worldwide there are millions of materials available commercially that are characterised by hundreds of different properties. Using traditional techniques to explore the information we know about these materials, to come up with new substances, substrates and systems, is a painstaking process that can take months if not years. Learning the underlying correlations in existing materials data, to estimate missing properties, the Alchemite
engine can quickly, efficiently and accurately propose new materials with target properties – speeding up the development process,” said Gareth, who is also a Royal Society University Research Fellow at the University of Cambridge. “The potential for this technology in the field of direct laser deposition and across the wider materials sector is huge – particularly in fields such as 3D printing, where new materials are needed to work with completely different production processes.”
Direct laser deposition – a form of DED – is used in many industries to repair and manufacture bespoke and high-value parts, such as turbine blades, oil drilling tools, and aerospace engine components, like the Stone Group is working on. As with most 3D printing methods, direct laser deposition can help component manufacturers save a lot of time and money, but next generation materials that can accommodate high stress gradients and temperature are needed to help bring the process to its full potential.
When it comes to developing new materials with more traditional methods of research, a lot of expensive and time-consuming trial and error can occur, and the process becomes even more difficult when it comes to designing new alloys for direct laser deposition. As of right now, this AM method has only been applied to about ten nickel-alloy compositions, which really limits how much data is available to use for future research. But Intellegens’ Alchemite engine helped the team get around this, and complete the material selection process more quickly.
(a) Secondary electron micrograph image for AlloyDLD. (b) Representative geometry of a sample combustor manufactured by direct laser deposition. [Image: Intelligens]
Because Alchemite can learn from data that’s only 0.05% complete, the researchers were able to confirm potential new alloy properties and predict with higher accuracy how they would function in the real world. Once they used the engine to find the best alloy, the team completed a series of experiments to confirm its physical properties, such as fatigue life, density, phase stability, creep resistance, oxidation, and resistance to thermal stresses. The results of these experiments showed that the new nickel-based alloy was much better suited for direct laser deposition 3D printing, and making jet engine components, than other commercially available alloys.
Discuss this story and other 3D printing topics at 3DPrintBoard.com or share your thoughts in the Facebook comments below.
We’re seeing an increased focus by software firms in our teeny tiny industry. Simulation, PLM, CAD, the Digital Twin, Industry 4.0 have some very large and influential firms salivating about that peanut butter and jelly sandwich that is the combination of the digital and manufacturing. If we are to grow digital manufacturing and 3D printing then our soft and hard assets will have to be managed and controlled through software. The more conventionally manufactured parts we replace with 3D printing the more files, settings, and process information will have to be monitored and accessed through software packages. This in part explains Dan Wellers’ interest in 3D printing and enthusiasm around the technology. He leads Digital Futures at SAP Global Marketing where in effect he has to be SAP’s Nostradamus and predict the impacts of technologies such as ours.
What is 4D printing?
Building upon existing 3D printing technology, 4D printing uses dynamic materials that perform differently when they encounter changing conditions such as water, light, heat, or electrical current. These materials—hydrogels, shape memory polymers, carbon fiber, custom textile composites, and more – have properties that enable objects to self-assemble, reshape themselves, or otherwise react to changing events or conditions. It’s called 4D printing because it incorporates what’s often referred to as the fourth dimension: time.
Why is it important?
4D printing can expand what is currently achievable in prototyping, design, manufacturing, and post-production adaptability and usage. Examples include: self-flattening boxes to be used in warehouses and logistics companies; plumbing system pipes capable of changing their diameter in response to flow rate or water demand. 4D printing has opened up entirely new innovations, such as medical implants made of dynamic biomaterials, which are already saving lives.
Because of its self-assembling capability, objects too big to be printed via conventional 3D printers can be compressed for printing and then expand after manufacturing. 4D printing could also be used to eliminate the mundane problem of furniture assembly. In addition, researchers have demonstrated how smart materials used in 4D printing can enable an object to “remember” its shape. That capability could be used to flat-pack a self-assembling shelter that springs into place after a natural disaster, or develop bridges and temporary roads made from materials that expand to heal damage and cracks.
How important will 3D printing be?
3D printing represents the digital transformation of both design and production in the manufacturing industry and will have a profound impact on everything from logistics to extended supply chains to trucking. 3D printing eliminates many of the design and production constraints inherent in traditional manufacturing. Product design can now be optimized for customer need and function, instead of production efficiencies.
Over the last few years, 3D printing has advanced in the way it employs different materials—not only plastic, but also metal, resins, sandstone, wax, and ceramics—increasingly incorporating multiple materials at once. These improvements are paving the way for significant benefits, including cost reductions, streamlined supply chains, faster time to market, increased personalization, optimized resource usage, improved prototyping, and the manufacture of new designs not possible in the past.
Commercial applications continually emerge to power 3D printing of everything from everyday household products to customized medical devices and prosthetics to nearly all the components of a house. 3D printing will expand into 4D printing in the coming years by adding a fourth dimension: time. By using specially engineered materials that perform differently when they encounter changing conditions, 4D printing promises to further shift the shape of manufacturing.
How will people and machines work together in the future?
Artificial intelligence is getting better at solving increasingly complex problems. If we want to retain humanity’s value in an increasingly automated world, we need to start recognizing and nurturing skills that are uniquely human.
To learn more, read “The Human Factor In An AI Future” and “Human Skills for the Digital Future”.
