Show Notes 26 April 2024
Story 1: Engineering household robots to have a little common sense – With help from a large language model, MIT engineers enabled robots to self-correct after missteps and carry on with their chores.
Source: MIT News Story by Jennifer Chu
Link: https://news.mit.edu/2024/engineering-household-robots-have-little-common-sense-0325
- From wiping up spills to serving up food, robots are being taught to carry out increasingly complicated household tasks. Many such home-bot trainees are learning through imitation; they are programmed to copy the motions that a human physically guides them through.
- It turns out that robots are excellent mimics. But unless engineers also program them to adjust to every possible bump and nudge, robots don’t necessarily know how to handle these situations, short of starting their task from the top.
- Now MIT engineers are aiming to give robots a bit of common sense when faced with situations that push them off their trained path. They’ve developed a method that connects robot motion data with the “common sense knowledge” of large language models, or LLMs.
- Side note, what are large language models? A large language model is a category of foundation models trained on immense amounts of data, enabling them to understand and generate natural language and other types of content for a wide range of tasks.
- These models are designed to understand and generate text like a human, infer context, provide coherent responses, translate languages, summarize text, answer questions, and even assist in creative writing or code generation tasks. Source: https://www.ibm.com/topics/large-language-models
- Here’s how MIT is working to give robots a bit of common sense – MIT’s approach enables a robot to logically parse many given household tasks into subtasks, and to physically adjust to disruptions within a subtask so that the robot can move on without having to go back and start a task from scratch — and without engineers having to explicitly program fixes for every possible failure along the way.
- A graduate student in MIT’s Department of Electrical Engineering and Computer Science noted, “With our method, a robot can self-correct execution errors and improve overall task success.”
- To illustrate their new approach the MIT team selected a simple chore: scooping marbles from one bowl and pouring them into another. To accomplish this task, engineers would typically move a robot through the motions of scooping and pouring — all in one fluid trajectory. They might do this multiple times, to give the robot a number of human demonstrations to mimic.
- The team then let the robot carry out the scooping task on its own…. As the robot moved through the steps of the task, the experimenters pushed and nudged the bot off its path and knocked marbles off its spoon at various points.
- Rather than stop and start from the beginning again, or continue blindly with no marbles on its spoon, the bot was able to self-correct, and completed each subtask before moving on to the next. (For instance, it would make sure that it successfully scooped marbles before transporting them to the empty bowl.)
Story 2: Intel unveils largest-ever AI ‘neuromorphic computer’ that mimics the human brain.
Source: LiveScience.com Story by Keumars Afifi-Sabet
- Scientists at Intel have built the world’s largest neuromorphic computer, or one designed and structured to mimic the human brain. The company hopes it will support future artificial intelligence research.
- Side note: Neuromorphic engineering is an interdisciplinary field that draws inspiration from biology, physics, mathematics, computer science, and electronic engineering. It aims to design artificial neural systems based on the structure and functioning of the human brain and nervous system. Source: https://en.wikipedia.org/wiki/Neuromorphic_engineering
- Intel’s machine, dubbed “Hala Point,” can perform AI workloads 50 times faster and use 100 times less energy than conventional computing systems that use central processing units (CPUs) and graphics processing units (GPUs), Intel said in a statement.
- Hala Point will initially be deployed at Sandia National Laboratories in New Mexico, where scientists will use it to tackle problems in device physics, computing architecture and computer science.
- Powered by 1,152 of Intel’s new Loihi 2 processors — a neuromorphic research chip — this large-scale system comprises 1.15 billion artificial neurons and 128 billion artificial synapses distributed over 140,544 processing cores.
- Side note: Synapses are crucial junctions in the nervous system where neurons (nerve cells) communicate with each other or with target cells like muscles or glands.
- Hala Point can make 20 quadrillion operations per second — or 20 petaops.
- Side note – Intel chose the name Loihi for its neuromorphic research processor with a purpose. The name Loihi is inspired by the Lo’ihi Seamount, an underwater volcano located off the coast of Hawaii.
- Optional side note – Here’s more on why Intel chose this name:
- Biological Inspiration: The field of neuromorphic computing aims to mimic the efficiency and performance of biological neural networks found in the brain. The Lo’ihi Seamount represents a natural phenomenon that shares similarities with the brain’s neural circuits.
- Parallel to Brain Function: Just as the Lo’ihi Seamount gradually rises from the ocean floor, Intel’s Loihi chip rises to the challenge of creating brain-inspired computing. The name symbolizes the gradual ascent toward achieving intelligent information processing at unprecedented efficiency levels.
- Metaphorical Significance: Naming the chip after an underwater volcano suggests a hidden power waiting to emerge. Similarly, Loihi embodies the untapped potential of neuromorphic computing, waiting to erupt with groundbreaking capabilities.
- In summary, Intel’s choice of the name Loihi reflects its commitment to unlocking the potential of neuromorphic computing, drawing inspiration from the natural world and the promise of brain-like intelligence.
- Let’s put 20 quadrillion operations per second into perspective — In numerical terms, a quadrillion is equivalent to 1 million million.
- In classical computing, binary bits of 1s and 0s flow into hardware like the CPU, GPU or memory before processing calculations in sequence and spitting out a binary output.
- WARNING, THE FOLLOWING IS SUPER GEEKY! In neuromorphic computing, however, a “spike input” — a set of discrete electrical signals — is fed into the spiking neural networks (SNNs), represented by the processors. Where software-based neural networks are a collection of machine learning algorithms arranged to mimic the human brain, spiking neural networks are a physical embodiment of how that information is transmitted. It allows for parallel processing and spike outputs are measured following calculations.
- Like the brain, Hala Point and the Loihi 2 processors use these spiking neural networks, where different nodes are connected and information is processed at different layers, similar to neurons in the brain. The chips also integrate memory and computing power in one place.
- In conventional computers, processing power and memory are separated; this creates a bottleneck as data must physically travel between these components. Integrating memory and computing power enables parallel processing and reduces power consumption.
Story 3: Scientists develop new material that could allow rechargeable batteries to ‘self-heal’ – The new cathode material for lithium-sulfur batteries is healable and highly conductive.
Source: The Cool Down Story by Rick Kazmer
- First, we need to set the stage with some battery basics –
- In the context of batteries, electrodes play a critical role in facilitating chemical reactions that generate electrical energy. For example, the cathode and anode in a battery are both types of electrodes. The cathode accepts electrons, while the anode releases them. Source: https://www.merriam-webster.com/dictionary/electrode
- What we’re going to be talking about here is the development of a lithium-sulfur cathode [electrode] with self-healing properties!
- Now for the news – A battery cathode in development in labs at the University of California San Diego has Wolverine-like self-healing properties.
- The regenerative ability of the newly developed lithium-sulfur electrode could help to unlock chemistry that doubles electric vehicle range, according to the experts. It’s a promising breakthrough with fascinating potential.
- Lithium-sulfur battery tech is already being developed in other labs, offering a clean, lightweight, and lower-cost alternative to common lithium-ion batteries.
- Side note – To learn more about lithium-sulfur batteries, see: https://www.greenbiz.com/article/promise-lithium-sulfur-battery
- UC San Diego experts take it to another level [as compared to other research and development to date on lithium-sulfur batteries] with a sulfur cathode that has a rejuvenation ability seemingly fit for the Marvel Universe.
- Here’s the problem – Cathodes in today’s lithium-ion batteries suffer structural damage, which accumulates during operation. To tackle this problem, the team made a cathode crystal from sulfur and iodine.
- In addition to self-healing of the new cathode [more on that in a moment] the iodine molecules have great impact, increasing conductivity “by 11 orders of magnitude, making it 100 billion times more conductive than crystals made of sulfur alone,” per the lab report.
- One of the researchers noted, “The drastic increase in electrical conductivity in sulfur is a surprise and scientifically very interesting.”
- Now, here’s more on the “self-healing” part of this discovery – Furthermore, the newly developed cathode material can melt at a low temperature — about 149 degrees Fahrenheit — at what the UC San Diego team likens to a cup of hot coffee. The low melting point allows the cathode to essentially heal damage during operation.
- One of the co-authors of the study noted, “Iodine disrupts the intermolecular bonds holding sulfur molecules together by just the right amount to lower its melting point to above room temperature yet low enough for the cathode to be periodically re-healed via melting.”
- It’s science that has great implications for battery longevity.
- A test battery maintained 87% capacity after 400 cycles. And the experts said this type of power pack can hold double the energy of lithium-ion batteries. That corresponds to twice the range for EVs, the researchers report.
Story 4: Engineer plans solar panel implant for human retina to retain eyesight.
Source: InterestingEngineering.com Story by Ameya Paleja
Link: https://interestingengineering.com/innovation/solar-panels-restore-vision
- Researchers at the University of New South Wales (UNSW) in Australia are working on a new prototype device that can be implanted on the eye’s retina to restore sight. The device uses solar panels like those used to convert sunlight into electricity, a university press release said.
- While a prosthesis replaces a missing or non-functional part of the body, neuroprosthesis is a relatively new field where the prosthetic device interacts with the neural system to restore lost functionality.
- Cochlear implants are a very good example of neuroprosthesis. In individuals with severe hearing loss, the implant converts sound into electric signals that are then used to stimulate the auditory nerve, restoring hearing.
- A similar approach could also be used for vision loss. “People with certain diseases like retinitis pigmentosa and age-related macular degeneration slowly lose their eyesight as photoreceptors at the center of the eye degenerate,” according to an engineer with expertise in photovoltaics at the University of New South Wales.
- The research team is using the same concept as cochlear implants to treat vision loss with a neuroprosthesis. For patients who have damaged photoreceptors the prosthesis must bypass them and convert light entering the eye into electric signals. These can then be directed to the optic nerve, enabling vision.
- To date, related research efforts have tried to create new photoreceptor cells, which involved experiments with electrodes that create a voltage pulse. While this works in principle, this system approach requires wires to go into the eye, making it a complicated procedure.
- The goal here is to find a no wires alternative – a University of New South Wales multidisciplinary team of researchers that includes engineers, neuroscientists, and biotech experts are working on achieving this with the help of solar panels. The idea involves using a tiny solar panel attached to the eyeball and converting incoming light into an electrical impulse sent to the brain.
- The major advantage of using solar panels instead of electrodes is that the device can be self-powered and does not require any wires to be attached to the eye. This isn’t the first time the idea has been experimented on, but the University of New South Wales team has taken a slightly different route toward achieving it.
- While solar panels are typically made of abundantly available silicon, the team has turned to alternative semiconductor materials such as gallium arsenide and gallium indium phosphide to develop his solar panels. Although more expensive than silicon, these materials are easier to work with, and their properties can be quickly modified to suit the application.
- To be ready for human implantation, the device must cover an area of two square millimeters, with each pixel measuring 50 micrometers. The University of New South Wales team added in the press release that users might wear special glasses or goggles that amplify sunlight to an intensity that can stimulate neurons in the eye.
Honorable Mentions:
Story: AI Scientists Create Humanoid Robot That ‘Thinks’ Its Way Through Tasks
Source: PC Magazine Story by Kate Irwin
Link: https://www.pcmag.com/news/ai-scientists-humanoid-robot-that-thinks-its-way-through-tasks
See video here: https://www.youtube.com/watch?v=Z3T9S1Arbdk
- Israel Tech startup Mentee Robotics has unveiled a prototype of its flagship robot that can listen to voice commands to complete a range of different tasks.
- The MenteeBot uses large language AI models (LLMs) to “think” through tasks from start to finish, making its own decisions along the way. The robot is 5’8″ in height, weighs about 154 pounds, and is expected to be released sometime in early 2025.
- “We are on the cusp of a convergence of computer vision, natural language understanding, strong and detailed simulators, and methodologies on and for transferring from simulation to the real world,” Mentee Robotics Chairman Asmnon Shashua said in a statement. “At Mentee Robotics we see this convergence as the starting point for designing the future general-purpose bi-pedal robot that can move everywhere (as a human) with the brains to perform household tasks and learn through imitation tasks it was not previously trained for.”
- Multiple videos show MenteeBot at work. In one, the robot determines the location of a kitchen table in an office environment, places fruit in a box without damaging it, and moves the box to a specific location.
- MenteeBot has a “voice” that the robot uses to communicate when tasks are nearly complete or to affirm that it’s heard the task. It’s able to navigate its environments without them being pre-programmed, as MenteeBot uses algorithms to map out the 3D physical space around it in real time, determines its own relative location, and is able to avoid obstacles as a result.
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
Story: The U.S. is exploring a railroad for the moon.
Source: Mashable.com Story by Mark Kaufman
Link: https://mashable.com/article/moon-lunar-railroad
- The Defense Advanced Research Projects Agency, or DARPA — an ambitious federal innovations division — has begun collaborating with over a dozen companies on potential future lunar technologies, including a moon railroad. It’s called the 10-Year Lunar Architecture Capability Study, or LunA-10, and its mission is to find technologies that will catalyze a self-perpetuating lunar economy. It’s a salient time; already the new space race is on.
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
Story: Intel bets its future on exotic, expensive tool that shoots lasers hotter than the sun – Late to the last generation of manufacturing technology, Intel wants to be out in front on the next.
Source: The Oregonian newspaper Story by Mike Rogoway
- At Intel’s massive Hillsboro research factory, another accessory has been commonplace lately: hard hats.
- Intel factory workers have spent months assembling an enormous manufacturing tool that weighs 150 metric tons, as much as a blue whale. It’s the most advanced lithography tool in existence, one of only two in the world, made by the Dutch company ASML.
- The giant piece of equipment is called High NA EUV. It is the size of a double-decker bus on the factory floor, but what you see is only part of the whole. There tool extends above the ceiling, and underneath the floor.
- From there, the EUV — extreme ultraviolet — tool shoots lasers at minuscule droplets of tin, firing 50,000 times a second with a burst of energy 40 times hotter than the surface of the sun. The resulting collision produces a shade of ultraviolet light that doesn’t occur naturally on Earth.
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
Story: Hydrogen cars set new range records
Source: MSN Story by “Gadget”
Link: https://www.msn.com/en-us/autos/news/hydrogen-cars-set-new-range-records/ar-AA1nxxcX
- At the end of February 2024, the UK ran its second Hydrogen Week. During this week, Hyundai set two new records for the driving range of its fuel cell electric vehicles (FCEVs).
- While certainly a significant achievement, will the ultimate range of FCEVs impact consumer opinions and adoption?
- The first record set was for the longest journey in the UK on a single tank of hydrogen at 406 miles using an ix35 FC. The second was the longest continuous journey in a hydrogen car, covering 6,000 miles over 6 days around the M25 in London. It should be noted there are a few other hydrogen cars that have stated ranges above this figure. However, being a fairly typical SUV, the ix35 is not a vehicle designed for the ultimate driving range at all costs but a realistic vehicle for private ownership.