Weekly Piece of Future #174
From Electronic Noses to Light-Based Computing and Robotic Strawberry Pickers
Hey there, fellow future-addicts!
Welcome to this week's edition of Rushing Robotics! This week's lineup reads like a wishlist from a decade ago: electronic noses that can sniff out a single walnut fragment, quantum computers hitting fidelity numbers that make fault tolerance feel imminent, gene therapies rewiring brain development in mice, and robotic strawberry pickers that know exactly when fruit is ripe. We're watching multiple frontiers—biotech, quantum, photonics, robotics, and AI—advance simultaneously, and the cross-pollination between them is where things get genuinely interesting. Whether you're here for the deep tech, the product launches, or just to marvel at what's possible, you're in the right place.
Unlock truly uncensored AI—privately.
While most AI platforms store and analyze your conversations on their servers, Venice keeps everything local to your browser—meaning your private thoughts and queries stay truly private with no server-side storage or data harvesting.
Venice leverages Trusted Execution Environments (TEEs) for end-to-end encrypted processing—ensuring that even during computation, your data remains cryptographically protected from prying eyes.
Get a $10 credit when you register through my link. Upgrade to Venice Pro and apply the credit toward the most advanced models immediately. No filters, no logs—just pure creative freedom.
Claim your $10 credit & start creating!
[Disclosure: I earn a commission if you sign up via this link.]
🤯 Mind-Blowing
From Berkeley's carbon-nanotube e-nose detecting trace allergens, to China's first dedicated photonic computing lab, to Harvard's electrically-driven DNA synthesis chip cranking out 64 sequences in parallel—this section is all about breakthroughs that reframe what's technically feasible. Add in Quantinuum's 99.9975% one-qubit fidelity and WVU's soft robotic fruit harvester, and you've got a snapshot of how fast the "impossible" becomes "inevitable."
🔊 Industry Insights & Updates
Manufacturing and robotics take center stage this week. Oak Ridge's origami-inspired 3D printing slashes fabrication time by 95%. ABB and PSYONIC are teaching robots dexterity through prosthetic-hand data. Alibaba's Qwen-Robot suite bridges language models with physical action. And Divergent Technologies is scaling metal 3D printing for defense at unprecedented throughput. The common thread? Industrial-grade systems are getting faster, smarter, and more adaptable—fast.
🧬 BioTech
Three stories, three different angles on fighting disease. UNIFESP identifies SDC4 as a metastasis shield worth targeting. UC Riverside restores normal brain activity in Fragile X mice via gene therapy. And TU Munich's CellTrap chip lets researchers watch single immune cells hunt cancer in real time. Together, they hint at a future where treatments are both more precise and more personalized.
💡 Products/Tools of the Week
From Rendervi's AI render studio for architects, to Lium's conversational AI for messy real-world datasets, to NoteRich's privacy-first note-taking, to CC Switch's multi-provider AI coding CLI manager—this week's tools are about giving professionals more leverage without sacrificing control.
🎥 Video Section
Genesis AI introduces Eno, Faraday Future unveils its FF EAI robot ecosystem, and Boston Dynamics takes us inside the lab to see how Atlas learns. Three very different visions of embodied AI—worth watching back-to-back.
When photonic chips accelerate the AI that trains the robots that assemble the gene therapy delivery systems, we're not just progressing linearly; we're building compounding loops of innovation. The next decade will be defined less by what any one technology can do alone, and more by what happens when they start working together. Stay hungry, stay futurish!
🤯 Mind-Blowing
Food safety monitoring could become more objective through an electronic nose created by researchers at the University of California, Berkeley, led by senior author Ali Javey and lead author Carla Bassil. The device uses 16 gas-sensitive carbon nanotube materials that react to different airborne molecules, addressing a longstanding manufacturing challenge where producing multiple sensing materials on a single chip has remained difficult. Carbon nanotubes form extremely thin conductive layers with large surface areas, making them highly sensitive to chemical compounds at room temperature and allowing the use of polymers that might degrade under heat. The system successfully identified as little as 0.05 grams of walnut material, roughly one-hundredth of an average shelled walnut, demonstrating potential for people with severe allergies. Bassil explained that the system uses the relative selectivity of gas sensors paired with machine learning pattern recognition to identify food-related chemical fingerprints more consistently than human smell alone, and she has already developed a portable version connecting to an iPhone application.
A new research laboratory dedicated to photonic computing has been launched in Shanghai by China, marking the country’s first industry-academia platform focused specifically on light-based chip technology. The Shanghai Key Laboratory of Integrated Photonic Computing Chips and Systems, based at Shanghai Jiao Tong University, will bring together researchers and industry partners to develop photonic chips, optical components, computing architectures, and software systems. The facility is a joint initiative between Shanghai Jiao Tong University and Lightelligence, a Shanghai-based startup that has claimed to be the first company to achieve large-scale deployment of hybrid optical-electronic computing systems. Zou Weiwen, director of the new laboratory and a professor at Shanghai Jiao Tong University, described photonic computing as an important pathway for achieving breakthroughs in computing power, offering advantages in bandwidth, latency, and energy efficiency.
A semiconductor chip capable of synthesizing 64 different DNA sequences in parallel using electric currents and a water-based enzymatic process has been developed by researchers at Harvard University, potentially offering an alternative to conventional DNA manufacturing methods. The chip uses localized electrical control to trigger DNA synthesis at selected sites on its surface, avoiding the solvent-heavy phosphoramidite chemistry widely used to produce synthetic DNA today. Each of the 64 synthesis sites contains two concentric ring electrodes surrounding DNA strands anchored at the center, with the inner electrode generating protons that lower pH to enable enzymatic DNA growth while the outer electrode consumes diffusing protons to prevent interference with neighboring sites. The approach generated 64 distinct DNA sequences simultaneously, with each sequence reaching up to 39 nucleotides in length, far exceeding the dozen sequences achievable with previous enzymatic methods.
Fault-tolerant quantum computing has moved closer to reality after Quantinuum's Helios quantum computer achieved record fidelity levels of 99.9975 percent for one-qubit operations and 99.921 percent for two-qubit operations. The 98-qubit system with 50 logical qubits uses trapped-ion qubits combined with photonics, reducing risks to quantum computing data while significantly decreasing power consumption to as little as 40 kW. Quantinuum is working with Sandia National Laboratories under a Cooperative Research and Development Agreement to develop fault-tolerant computers, with SNL researchers developing a new benchmarking methodology that measures the performance of mid-circuit measurements, non-destructive readout operations that help correct quantum computing errors. Robin Blume-Kohout, a researcher at SNL involved in the work, said the laboratory evaluates every aspect of quantum computer performance with commercial partners to accelerate the advent of quantum supercomputing.
A soft robotic gripper capable of detecting fruit ripeness and harvesting without damage has been developed by researchers at West Virginia University, led by assistant professor Anand Mishra from the Department of Mechanical, Materials and Aerospace Engineering. The device uses soft silicone and polyurethane fingers equipped with multiple sensors to measure a fruit’s size, shape, color and firmness before deciding whether it is ready to be picked. Tested on strawberries, the system removes fruit by twisting its stem rather than cutting it, addressing a major challenge in agriculture where delicate fruits have narrow harvest windows and bruise easily during picking, transportation and storage.
🔊 Industry Insights & Updates
Fabrication time has been cut by 95 percent and costs reduced by 90 percent through a new origami-inspired 3D printing method developed by Steven Guzorek and fellow researchers at Oak Ridge National Laboratory. The mold-free process uses a high-strength fabric substrate such as nylon, glass fiber, or resin-infused fibers, combined with an integration layer of thermoplastic polyurethane to ensure adhesion. Reinforcing materials including thermoplastic carbon-fiber acrylonitrile butadiene styrene for lightweight strength or thermoset resins like epoxy for greater stiffness are then deposited, forming molecular-level bonds that create a robust, unified structure. Guzorek explained that the pioneering method redefines advanced manufacturing by fusing material science with transformative design principles, improving the efficiency and scalability of large-structure manufacturing while achieving forms unattainable with traditional additive approaches.
A partnership has been formed between ABB Robotics and California-based bionic company PSYONIC to develop more dexterous robotic systems using real-world data from human prosthetic use. The collaboration combines PSYONIC's Ability Hand, a touch-sensitive prosthetic hand, with ABB's GoFa collaborative robot to study how human-generated touch and motion data can train robots to perform delicate and variable tasks. The initiative aims to address one of industrial robotics' biggest challenges—grasping and manipulation—while improving automation capabilities and potentially reducing engineering time for handling applications by up to 30 percent. Marc Segura, President of ABB Robotics, noted that human dexterity and the instinctive understanding of how to handle different objects remains one of the most difficult things to replicate in industrial-grade robotics.
A first-of-its-kind embodied AI model family linking large language models with real-world robotic actions has been launched by Chinese firm Alibaba through its Tongyi Lab. The Qwen-Robot suite, currently undergoing pilot testing with selected Alibaba Cloud enterprise clients, comprises three specialized models targeting different layers of physical intelligence. Qwen-RobotNav focuses on movement and navigation, helping robots follow instructions, navigate to locations, track targets, and support autonomous driving. Qwen-RobotManip enables physical interaction such as grasping, moving, and manipulating objects using training data collected from different robotic systems. Qwen-RobotWorld acts as a world model that predicts how environments may change and helps robots understand likely outcomes of their actions, with the combined suite enabling machines to perceive, reason, and interact with the real world.
Production throughput could roughly double compared with current-generation equipment through the Monolith One metal 3D printer developed by California-based Divergent Technologies. The machine offers a larger build chamber than many existing systems, allowing engineers to manufacture bigger and more complex structures in a single print. Divergent currently operates six Monolith One printers at its headquarters in Torrance, California, and plans to install an additional 64 systems at a new 430,000-square-foot facility in Long Beach over the next two years. Once fully operational, the site will support production of tens of thousands of munition airframes annually and have capacity to manufacture hundreds of thousands of critical metal components for aerospace and defense programs. The expansion comes as defense contractors seek faster production methods and more resilient domestic supply chains, with traditional manufacturing processes often requiring lengthy lead times for complex hardware.
🧬 BioTech
A molecule on the surface of cells that could serve as a promising therapeutic target against cancer has been identified by researchers at the Federal University of São Paulo (UNIFESP) in Brazil. The study, led by full professor Carla Cristina Lopes in the Department of Biological Sciences and supported by FAPESP, showed that overexpression of the Sindecam-4 (SDC4) receptor provides tumor cells with a protective shield against anoikis, a type of programmed cell death that occurs when cells detach from tissue. When researchers silenced the receptor using genetic engineering techniques, tumor cells lost their malignant properties and became dependent on adhesion to survive once again. The findings, published in the journal Cytotechnology, suggest that SDC4 could become both a therapeutic target for stopping metastasis before it begins and a diagnostic marker for monitoring disease progression.
A gene therapy designed to replace a missing brain protein restored normal brain activity and improved behavior in a mouse model of Fragile X syndrome, according to a University of California, Riverside-led study published in Molecular Therapy Nucleic Acids. The research, led by senior author Iryna Ethell, a professor of biomedical sciences in the UCR School of Medicine, used a modified adeno-associated virus called AAV9 to deliver a healthy human version of the FMR1 gene into newborn mice lacking FMRP, the protein that regulates communication between brain cells. High-dose treatment produced significant improvements including normalized gamma brain-wave activity, reduced background neural noise, improved responses to sound, normal exploratory behavior, stronger social interactions, and improved cognitive flexibility. First author Anna Norman, a project scientist in Ethell's lab, and the research team found that timing was critical, with therapy administered during an early developmental period when the brain remains highly adaptable. The findings suggest gene therapy may one day address the underlying cause of FXS rather than simply treating its symptoms.
Detailed observation of how individual immune cells interact with cancer cells has become possible through CellTrap, a microfluidic chip developed by Ghulam Destgeer and his research team at the Technical University of Munich's School of Computation, Information and Technology. The platform features 1024 small trapping chambers at the ends of continuously branching pathways from a main channel, into which cells are drawn and spatially fixed for observation over many hours. Unlike established laboratory tests that mainly capture average values across many cells, CellTrap reveals how each cell reacts and interacts with others, providing crucial timing information about when contact, activation, and killing of cancer cells occur. The researchers tested the platform on three cancer cell lines including glioblastoma, chronic myeloid leukemia, and adenocarcinoma, demonstrating that early activation signals in immune cells often predict later cell-damaging effects within the same interaction. Destgeer noted that the platform is not limited to immune and cancer cells, as almost any combination of cells can be loaded and observed in the chip.
💡Products/tools of the week
Rendervi, an AI render studio, converts model previews, sketches, and clay renders into consistent, client-ready photorealistic images and short videos. The platform leverages image-to-image AI technology, currently utilizing Nano-Banana Pro, and integrates background models such as ChatGPT and Gemini to analyze visuals and build optimized prompts. Rendervi equips architecture teams with controlled inpainting tools, material swap options from a comprehensive texture library, environment tweak capabilities, and presets designed to maintain cross-view consistency across renderings. With one-click upscaling functionality included, the studio enables architects to iterate faster while preserving their design intent and significantly reducing manual rework throughout the creative process.
Lium – AI for Complex Data, a conversational AI platform and infrastructure solution, connects, indexes, and profiles messy, terabyte-scale real-world datasets spanning geospatial, energy, space, infrastructure, and related fields. The platform automatically generates code, tools, and data transformations, provisions compute resources on demand, and allows teams to query and produce reusable analyses and outputs using natural language. Lium empowers domain experts to obtain fast, reliable insights without requiring heavy engineering involvement.
NoteRich keeps users' original notes 100% local while using ephemeral, minimal-context cloud calls to deliver AI features like summarization, smart Q&A, scan & extract, memory-driven personalized insights, and local RAG indexing. The app provides users with powerful, evolving AI assistance along with cross-device P2P sync and encrypted storage, all without permanently uploading or exposing their full note archive. NoteRich makes modern AI productivity accessible to users who demand strict data control.
CC Switch centralizes and visualizes configuration, provider keys, model routing, and failover for multiple AI coding CLIs including Claude, Gemini, Codex, Hermes, OpenCode, and OpenClaw, enabling developers to import presets, switch providers from the system tray, and avoid hand-editing JSON, TOML, or env files. The tool, developed by Jason Young (farion1231), runs a local API proxy, manages MCP servers and Skills, and tracks sessions and usage so comparing models, recovering from outages, and scaling multi-provider AI development becomes simpler and more reliable for developers and teams.
