Weekly Piece of Future #142
From Neuromorphic Chips to Bio‑Hybrid Robots and 4‑D Ultrasound Imaging
Hey there, fellow future-addicts!
Welcome to this week's edition of Rushing Robotics, where the impossible gets a headline and the frontier of intelligence is laid out front and center. In this issue we’ll walk you through the most groundbreaking stories in neural engineering, consumer robotics, and bio‑AI.
🤯 Mind-Blowing
Researchers at USC have created a diffusive memristor that truly mimics the electro‑chemical signals of living neurons, promising a new generation of ultra‑low‑power neuromorphic processors that learn on the fly. In a parallel leap, scientists have fused living muscle cells onto a synthetic scaffold, producing a bio‑hybrid chassis that contracts like muscle and could lead to robots that grow and repair themselves. Meanwhile, a protein‑language model developed in Glasgow now predicts protein folding and interactions even faster and more accurately than AlphaFold, opening a new era for drug discovery and synthetic biology.
🔊 Industry Insights & Updates
The first commercially available humanoid robot for home use has entered the market, offering companionship, chores, and a flexible platform for developers to build new applications. Princeton’s latest exascale supercomputer is now crunching climate and scientific data at a pace that was unimaginable a decade ago, while a partnership between national labs and industry has unveiled a quantum‑enhanced hydrogen‑production facility that slashes clean‑fuel costs by more than half. These breakthroughs illustrate how AI is rapidly moving from laboratory curiosity to tangible industry solutions.
🧬 BioTech
Glasgow’s PLM‑Interact outperforms AlphaFold by accurately predicting complex protein–protein interactions in real cellular environments, accelerating biomarker identification. Inserm researchers introduced a 4‑D ultrasound Doppler system that visualises blood flow from arteries to arterioles in real time, giving surgeons an unprecedented live map of organ circulation. At EPFL, a nanoscopic micro‑catheter capable of navigating vessels thinner than a human hair promises targeted drug delivery deep into the brain’s microvasculature.
💡 Products/Tools of the Week
Komos AI automates repetitive workflows by creating repeatable processes from a single screen demo, allowing teams to prototype and iterate faster. JoggAI turns text prompts into lifelike avatar videos, opening creative avenues for marketers and educators alike. GrapesJS offers a no‑code web‑design editor enhanced with AI helpers, so designers can build responsive sites without writing a single line of code. Liminary aggregates knowledge across PDFs, videos, and transcripts, turning scattered information into a single, searchable knowledge base that keeps your creative pipeline humming.
🎥 Video Section
In our latest video showcase, watch a new home‑robot perform a cleaning routine, play music, and interact with pets—demonstrating the strides being made in consumer robotics.
The pace at which AI is redefining the boundaries of biology, computation, and everyday life is nothing short of exhilarating. Each breakthrough feels like a stepping stone toward a future where machines not only process data but also grow, heal, and collaborate with us. Stay hungry, stay futurish!
🤯 Mind-Blowing
New artificial neurons mirror authentic brain chemistry, paving the way for smarter AI hardware. USC’s Viterbi School of Engineering and the School of Advanced Computing have fabricated neurons that physically emulate the electrochemical dynamics of living brain cells. This breakthrough marks a pivotal move toward highly efficient, brain‑like computing systems that might, in the future, underpin artificial general intelligence. Conventional neuromorphic chips approximate brain function through digital simulation, whereas USC’s neurons harness genuine chemical and electrical interactions to perform calculations.
In essence, they don’t merely imitate how the brain operates—they behave like actual biological neurons. The project was spearheaded by Joshua Yang, a Professor of Computer and Electrical Engineering and Director of USC’s Center of Excellence on Neuromorphic Computing. Yang’s team introduced a “diffusive memristor”—an artificial neuron that, unlike traditional silicon devices reliant on electron flow, exploits atomic diffusion to process data. By employing silver ions within an oxide matrix to emulate the brain’s electro‑chemical synaptic transmission, they integrated each neuron into a single transistor, a breakthrough that promises smaller, faster, and more power‑efficient AI hardware.
New research paper suggests robots that could live on muscle cells like humans, ditching gears and motors, with engineers and biologists fusing living tissue with synthetic structures to create humanoids and robots that behave more like human beings. The current experiment is a part of the emerging field known as ‘biohybrid robotics’, in which advanced fabrication technologies are used to build, control, and sustain such systems. If the research proves successful, the next generation of robots might flex, contract, and grow using living muscle; to achieve their goals, scientists use two types of muscles: first skeletal muscle, which moves when it receives an electric signal, and second cardiac muscle, which beats on its own to provide continuous, coordinated motion. Both muscles offer unique advantages but equally present a set of challenges, as muscle tissue is delicate, demanding, and short‑lived outside a human body, requiring nutrients, oxygen, and the right environment to survive; the research paper has highlighted four key methods—3D bioprinting, electrospinning, microfluidics, and self‑assembly—that allow scientists to arrange muscle cells precisely and nurture them within engineered scaffolds. With these techniques, cells can align, grow, and contract in unison, turning living tissue patches into functional actuators.
Scientists uncover a hidden ‘geometric code’ that helps DNA compute and remember, as a new Northwestern study reveals a second language of life—a geometric code that shapes how our genome thinks. Led by biomedical engineer Vadim Backman, the study shows that the three‑dimensional physical structure of DNA contains a “geometric code,” a system allowing cells to perform calculations, retain memories, and adapt. Inside each cell, DNA folds into nanoscale “packing domains” that act as physical memory nodes, units capable of storing and stabilizing genetic activity. These structures are not random; over millions of years, evolution appears to have refined the genome’s shape to increase its efficiency in storing and accessing information. Scientists believe this geometric language could be the bridge between biology and computation. Understanding this hidden code could open new ways to repair or rewrite cellular memories.
A self‑powered spinal implant capable of transmitting healing data from inside the body has been unveiled, marking the world’s first such device that also harvests its own power and tracks bone repair progress without external batteries.
First demonstrated in a series of trials, the battery‑free metamaterial implant not only powers itself but also continuously monitors the progress of bone healing by converting mechanical pressure into electrical signals that are sent back to clinicians. By weaving conductive and non‑conductive layers into a human‑made composite, the metamaterial design captures ambient pressure to both generate electricity and transmit diagnostic signals, ensuring continuous connectivity without radiation or wired links. Leveraging generative AI, engineers were able to produce patient‑specific metamaterial layouts for each spine, dramatically cutting development time and tailoring performance to individual anatomy while maintaining consistent signal fidelity. Immediately after surgery, the implant emits a stronger signal due to the increased pressure from the vertebral endplates on the cage, while the signal naturally diminishes as bone remodeling transfers more load to the healing bone, providing a real‑time indicator of successful recovery.World’s tiniest 3D bioprinter injects healing gels into damaged vocal cords, as researchers at McGill University have engineered a flexible printing device that can slide into the human throat to rebuild vocal folds. Targeting damaged tissue with hydrogels, the apparatus is only 2.7 mm wide, marking it as the smallest bioprinter yet documented. The paper explains how this surgeon‑controlled, bendable tool could restore patients’ voices after surgery by accurately regenerating fragile vocal tissues, surpassing previous methods in precision.
After vocal‑cord procedures, many patients suffer from rigid, scarred folds that impede speech. Roughly 3–9 % of individuals develop voice problems from cysts, tumors, or other growths that necessitate removal. While surgeons usually inject hydrogels to reduce scarring, delivering them precisely in the throat’s tight space is challenging. The team set out to build a miniature printer that would fit within a patient’s airway without obscuring the surgeon’s view, since existing bioprinters for organs like the liver or colon are too large for such a delicate setting.
🔊 Industry Insights & Updates
NEO, the first consumer‑ready humanoid that folds laundry, tidies rooms, and learns new skills over time, is slated to launch in 2026 in the United States, with preorders now open for early‑access units priced at $20,000 and a $499‑per‑month subscription plan to widen access. The machine promises to automate household chores and provide personal assistance at home, and robotics firm 1X has described it as the safest, most capable, and most affordable humanoid ever built, turning the sci‑fi fantasy of humanlike home robots into a reality that people can bring into their living rooms. NEO’s Chores feature lets owners assign, schedule, and track tasks using simple voice or app commands, allowing the robot to tidy rooms, fold laundry, clean up spaces in real time, and for unfamiliar jobs, customers can book a 1X Expert to guide the robot while it learns on the fly. Beyond its physical work, the humanoid is equipped with a built‑in large language model that supports conversation and contextual understanding without the need for screens, Audio Intelligence that ensures the robot listens only when addressed, and Visual Intelligence that enables it to identify objects and situations—spotting ingredients on a counter and suggesting recipes when it recognizes a cooking scenario. The robot’s Memory system provides continuity across conversations, remembering grocery lists, birthdays, or past discussions and adjusting its responses over time, making it not just a helper but a learning companion that grows smarter with each interaction; NEO will ship in Tan, Gray, and Dark Brown starting in 2026.
A collaboration between NVIDIA and Oracle, is poised to redefine scientific discovery by building the nation’s largest AI supercomputer. The new system will sit at Argonne National Laboratory and use NVIDIA’s high‑speed networking to deliver 2,200 exaflops of AI performance. The partnership will enable researchers to train cutting‑edge AI and reasoning models with NVIDIA’s Megatron‑Core library and TensorRT inference software, creating what NVIDIA calls “agentic AI workflows” that support open science. The supercomputer’s architecture also introduces NVQLink™, a novel open system that connects GPU clusters with quantum processors, ushering in an era of accelerated quantum‑classical computing. NVQLink, offers ultra‑fast, low‑latency data exchange between GPUs and quantum units, delivering the performance necessary for next‑generation hybrid workloads. The new infrastructure will accelerate innovation across scientific disciplines, from materials science to climate modeling, by providing unprecedented compute power for training and inference tasks. By combining NVIDIA’s AI expertise, Oracle’s enterprise solutions, and DOE’s research mandate, the joint effort aims to unlock transformative insights that were previously beyond reach, setting a new benchmark for large‑scale scientific computing.
Supramolecular robotics enables soft materials to move, adapt, and self‑assemble, as a research team from Japan has outlined a fresh framework that lets these materials display motion, transformation, and self‑assembly by actively tuning molecular interactions. While many bio‑inspired materials imitate particular biological functions, most are limited to a single stimulus and do not exhibit the integrated responsiveness of living systems; in this approach, molecules serve as adaptable building blocks that can organize, disassemble, and reorganize in response to subtle chemical cues, producing materials with programmable motion, shape transformation, and cooperative assembly—bridging the realms of molecular chemistry and robotic behavior. The investigators identified the three foundational principles of supramolecular robotics: motility, phase transition, and prototissue formation. Depending on the stimulus, droplets may move directionally or form collective patterns that resemble microbial swarms, and these chemically powered motion systems could lay the groundwork for microscale robots capable of environmental sensing or targeted transport.
Scientists at Princeton announced a breakthrough that turns wastewater into hydrogen fuel, offering a method that cuts the high cost and environmental impact associated with the clean water required for hydrogen production. In this advance, the research team demonstrated that reclaimed wastewater can replace ultrapure water in the electrolysis process that generates green hydrogen, thereby bypassing the expensive and environmentally burdensome water purification step. The study involved treating reclaimed wastewater—water that is suitable for non‑drinking uses such as irrigation or industrial cooling—with sulfuric acid, a step that acidifies the water and enriches it with protons.
The acidified buffer outcompetes the calcium and magnesium ions that normally interfere with electrolysis, ensuring that ion conductivity across the membrane remains high and that the electrical current is sustained. With these conditions met, the system can produce hydrogen continuously without failure, marking a significant step toward more sustainable and cost‑effective green hydrogen production.
🧬 BioTech
A supercomputer aided the creation of an artificial intelligence model that interprets the “language of proteins.” The University of Glasgow scientists utilized the Tursa supercomputer, part of the UK’s DiRAC High Performance Supercomputer facility, an engine usually dedicated to cosmic research but employed for medical research in this work. Tursa facilitated the development of the protein language model PLM‑Interact, which predicts protein interactions and can foresee mutations that will interrupt communication among these essential molecules. Initial trials reveal PLM‑Interact outperforms existing models, including Google DeepMind’s AlphaFold3, in forecasting protein interactions. In medical science, this advancement promises a superior method for comprehending disease mechanisms such as cancer and viral infections.
A new ultrasound‑based imaging technology was developed that can map organ blood flow in four dimensions—three spatial dimensions plus time—offering a level of detail previously unattainable. The technology, created by Inserm researchers at the Physics for Medicine Institute in France, could provide deeper insights into the circulatory system and enhance the diagnosis and treatment of blood‑circulation‑related diseases. For the first time, the technique mapped the 4D blood flow of entire organs in animal models, addressing a key medical challenge: the lack of a tool capable of visualizing the entire circulatory network within a whole organ, from large arteries to the smallest arterioles. The team’s non‑invasive ultrasound probe is the first to solve this problem, delivering unprecedented image resolution that enabled mapping of vascularization and precise quantification of blood‑flow dynamics in the heart, kidney, and liver in animal models comparable to human size. This non‑invasive tool can distinguish microcirculation in vessels under 100 micrometers and, if applied clinically, could become a major instrument for understanding vascular dynamics from the largest vessels to pre‑capillary arterioles.
A new microcatheter developed by scientists at the Swiss Federal Institute of Technology Lausanne (EPFL) can hitch a ride on the bloodstream to navigate the body’s narrowest arteries, even thinner than a human hair. Microcatheters are already critical for lifesaving procedures such as clearing clogged arteries, stopping bleeding, or delivering drugs directly to tumors. However, existing guidewire‑based systems are slow, difficult to steer, and can damage vessel walls — and they can’t reach the smallest, most tortuous vessels deep within the brain. The innovation could open new possibilities in treating conditions like stroke, arteriovenous malformations, and even eye cancers in children.
💡Products/tools of the week
Komos AI converts a short on‑screen demonstration into a fully functional, repeatable workflow that harnesses computer vision and artificial intelligence. It watches user actions and listens to voice narration, then automatically assembles a deterministic, editable process that can be executed safely and consistently. Users can visually tweak steps, insert logic such as loops and conditional branches, and securely store credentials using encrypted storage. The platform supports integrations with webhooks, REST APIs, emails, and spreadsheet exports, all designed for production‑grade reliability, eliminating repetitive manual tasks in operations like record updates, invoice handling, and data consolidation while preserving auditability and predictable outcomes.
JoggAI generates realistic avatar‑based videos without filming or editing, using AI to create lifelike avatars that deliver messages naturally for ads, social media, and other digital uses. It lets marketers, creators, and businesses produce high‑quality videos in minutes by turning text, photos, or custom avatars into fully produced content, complete with accurate lip‑syncing, AI voices, captions, and built‑in editing tools. The platform cuts production costs and time drastically, automating the entire video creation process to deliver professional, ready‑to‑publish material instantly.
GrapesJS is a free open source web template editor framework that lets users build web pages, landing pages, email templates, and newsletters through a visual drag‑and‑drop interface requiring no coding while its core features focus on flexible template design and clean code export. GrapesJS also now includes AI‑powered tools such as automated website generation, website cloning, and point‑and‑click content editing in its Studio Beta enabling users to create and modify designs even faster and more intuitively. Developers can embed GrapesJS in their own platforms, designers can rapidly prototype, and marketers or non‑technical users can craft campaigns all without technical barriers, making it an ideal choice for anyone seeking a customizable and extensible solution for professional‑quality web and email content production.
Liminary automatically captures and organizes content from articles, PDFs, videos, and meeting transcripts, allowing knowledge workers to instantly recall and synthesize information by surfacing the right insights exactly when they need them, thereby eliminating time wasted searching for information and preserving creative flow. The platform employs AI agents to surface relevant knowledge, map connections, and verify sources, features that prove especially valuable for marketers and strategy teams integrating external research and internal knowledge into campaign briefs, reports, and other work artifacts, all while maintaining a privacy‑first approach.
