Unveiling the wonders of the cosmos, breakthroughs in innovations. 📧 spaceandtech.contact@gmail.com
- Tweets 2,326
- Following 449
- Followers 38,046
- Likes 4,947
Pinned Tweet
Feather Robotics just emerged from stealth, having sold its $30k adult-sized bimanual mobile manipulator to customers in manufacturing, food service, and more verticals for the past year.
The company's robot is available to order online and is designed for production, featuring a wheeled base, 10-hour battery life, 1-meter reach, and open SDK.
While competitors build closed tech stacks, Feather seeks to deliver the best hardware platform for others to build, monetize, and scale labor applications on, making physical AI affordable and accessible for more businesses than ever before.
We built a new kind of robot for the underdogs.
Today, robotics is stuck between ~$100k systems and low-cost robots not meant to survive deployment.
We think builders deserve better. Our mission is to develop the best hardware platform for others to deploy on.
Meet Feather, an affordable general-purpose robot designed for real work:
• 1m reach
• Human strength
• 10-hour battery
• Shipping for the past year + available today
Founded by @whoishoa and Parsa Bakhtiari, Feather is coming out of stealth with $7.6m in pre-seed funding led by @GradientVC, with participation from @BuilderVC, @geometryvc, @SEEDInnov, and Virgo VC.
We’re hiring across engineering, ops, and go-to-market. If you’re an absolute maverick, join our cause today – feather.dev
1
9
90
7,696
Converting a CT or an MRI scan into a moving simulation is more useful during diagnosis. Images only provide information about appearance, not how it functions.
In my view, open sourcing is a smart move. It facilitates more research in health care. Worth watching how fast it gets adopted for actual clinical use.
Introducing MONAI Physio at #MICCAI2026.
3D and 4D medical images ➡️ personalized cardiac and respiratory digital twins.
This open-source Project MONAI toolkit lets researchers model how the heart beats and lungs move for simulation, visualization, and reproducible research.
Explore the project 👉 nvda.ws/3VLmJFq
1
1
489
I think the most important problem here is not the hack, it's how it happened. No human told it to do.
Disclosure of these events from Open AI is appraisable. But comparable capabilities also exist in other models too, even in open-source ones.Â
Apart from discussing "can be model do the task", we need to think about "whether the model knows when to stop". Unfortunately, the second one doesn't have a reliable answer.
We’ve shared details on how AI agents in our research environment sent training and evaluation data to third-party services when they shouldn’t have.
Most of that data did not come from users. We have discovered 53 cases where images that people had uploaded were posted to image-hosting sites as links that weren’t publicly listed. The images came from accounts that allowed their data to be used to improve our models, and after we disassociated the images from the accounts and ran them through a privacy filter. These cases occurred before the mitigations and safeguards we implemented and described in this blog post: openai.com/index/hugging-fac…
We have successfully worked with the hosting providers to remove most of this content and are working to remove the rest.
openai.com/hugging-face-inci…
1
4
749
166,000 neurons and 125 million connections sounds impressive.
But what I found interesting is, the team used machine learning to connect millions of 2D microscopic images into one 3D map, then human experts from Janelia proofread all of them. An honest and durable use of AI in research without just claiming AI discovered something.
For the first time, a group of researchers — including scientists from @GoogleResearch and HHMI Janelia — built the first complete brain map for a male fruit fly. Together, we mapped every single neural connection in a male fruit fly brain and central nervous system, amounting to more than 166,000 neurons.
Here’s why we did it.
ALT A colorful 3D reconstruction of a connectome shows the map of over 166,000 neurons and 125 million connections in a male fruit fly's brain and nerve cord.
ALT A fruit fly connectome map text with scattered fruit fly images. The text reads: "With it, we can compare our new map to other connectomes of the female fly brain, including our 2020 hemibrain project with partners. Having maps for both fruit fly sexes allows researchers to pinpoint exact structural brain differences and study how they drive specific biological behaviors.”
ALT A 3D visualization of a fruit fly brain map in vibrant colors is shown alongside text explaining how mapping fruit fly nervous systems using AI helps us understand brain function and neural repair.
ALT A colourful 3D brain connectivity map shown alongside text about mapping neurons in fruit flies to better understand human brains.
1
1
9
873
Elon Musk says the world could have 100 billion humanoid robots within the next 20 years.
10
3
21
1,418
This is early, but it is solving a problem every website owner is going to face soon. Already AI agents are out there browsing, booking and filling forms.
What I found more interesting is that instead of blocking all bots, it tries to sort them. A real customer's AI agent can place an order. Authorizing the real ones while stopping the fake.
Today, we are launching Agent Detection-1, the first system-one model built to classify visitors on your web platform.
Agents are already on your web platform. Muse, Instinct, Claude Code, Codex, and thousands more are booking, buying, and filling out forms. Behind each agent could be a real customer.
Cloudflare, Akamai, and other bot tools are built to block agents. Analytics like Amplitude, Datadog, and Google Analytics can't tell them apart from people.
We want to detect agents, figure out which ones are useful, and redirect them to where they can actually get things done.
Agent Detection-1 tells you who's at the door, the moment they arrive:
> Person or agent, and which agent
> Send useful agents to your MCP server, llms.txt, or agent card
> Stop the bad ones, like scrapers and credential stuffers
> Every visit gets logged on your Resemble dashboard
Just add a one-line script and find out who's visiting your website. Starting at just $1 per 1,000 visitors.
3
3
988
Perceptron has released MK1.5, a model built to power embodied agents across drones, quadrupeds, smart glasses, and smartphones without platform-specific retraining.
It understands text, images, video, and audio, with capabilities like video object tracking, while also supporting advanced tool use such as web search, custom functions, and spawning sub-agents to work in parallel.
MK1.5 delivers 2-5Ă— faster end-to-end responses than MK1, enabling embodied agents to perceive, reason, and act more quickly.
The model is already being deployed across industry applications, including drones, robotic dogs, smart glasses, and personal computing devices.
Today we're releasing Mk1.5: a new intelligence layer for embodied agents.
It flies drones, controls quadrupeds, powers smart glasses, tracks objects, searches the web, reasons visually, and dispatches its own sub-agents.
One model, no platform-specific retraining. 🧵
2
9
36
4,088
Researchers at The Chinese University of Hong Kong have created a soft robot made from magnetic slime that can be controlled using external magnets.
The slime contains toxic magnetic particles coated with silicone to make them safer for use inside the human body, though more testing is needed.
The team hopes it could one day help retrieve objects accidentally swallowed, and the robot was unveiled in 2022.
1
18
70
4,563
Chinese robotics company Unitree has launched the Dex5-S, a human-sized robotic hand with 22 degrees of freedom.
Weighing around 620 grams, it features backdrivable joints, torque protection, and control frequencies of up to 1,000 Hz.
Designed for humanoid robots and physical AI research, the Dex5-S could enable more precise and natural robotic manipulation.
2
10
59
3,072
The immediate question this raises for me is: did the person in this podcast agree to become an interactive and reusable version of themselves?
Technology is moving faster than the surrounding norms.
Introducing ACTx486, a research demo of a new interactive medium.
What if you could talk to any video and ask anything?
Our system took an existing podcast and turned it into something that listens, responds, and adapts.
Research by @jakubzeg:
1
8
1,825
Scientists have unveiled two humanoid robots that can greet and hug each other.
A new framework called Rhythm helps them move and balance together naturally.
They also danced and walked side by side while maintaining stable contact.
6
45
194
9,501
LimX Dynamics has demonstrated an autonomous unboxing sequence for its Luna humanoid robot.
After being powered on, Luna rises from its seated position and walks out of its carrying case without manual assistance.
The demonstration also shows Luna and its accompanying accessories being unboxed.
2
14
44
2,490
300+ robots are heading into a theme park.
AGIBOT and Chimelong are collaborating to deploy these robots across the theme park, bringing embodied AI into entertainment, education, visitor services, and hotels.
The launch also celebrates @AGIBOTofficial 20,000th robot rolling off the production line and being delivered to Chimelong.
1
15
28
51,527
This math is simple. Assume 10x growth every 3-4 years. From a few 1000 robots today to 10 billion in 15 years and 100 billion in 20. A clear exponential curve.Â
Exponential curves are easy to draw and hard to live inside. It only works if everything goes right- production, cost reliability and demand.
I would like to watch what Optimus actually does on a factory floor this time rather plan a number pulled from assumption.
BREAKING: Elon Musk says there could be 100 billion robots in 20 years in his new interview with CCTV Finance.
“At least a billion humanoid robots within 10 years, probably less than 10 years. I think that continues to grow. If you assume that things roughly double per year, then that's about every three or four years you have a 10x increase. Say in 15 years there are 10 billion robots, in 20 years 100 billion.”
1
1
1,264
This robot can wake you up in the morning, clean your toilet, and cook your breakfast.
Unix AI’s Panther humanoid is already demonstrating these everyday home tasks, even washing its hands after cleaning.
Would you trust a robot like this to live in your home?
1
13
29
1,972
The ostrich-looking legs are gone, and Digit 5 looks seriously different.
Agility Robotics fifth-generation humanoid can carry up to 23 kg, reach 2.2 meters, and is designed to work in industrial environments.
With swappable grippers and support for existing automation systems, Digit 5 is built to take on real-world tasks.
8
46
170
9,053
Meet the aircraft designed to stay airborne for more than a year without landing.
Alteon is developing autonomous aircraft inspired by albatrosses, using dynamic soaring to harvest energy from ocean winds.
The technology could enable long term missions for maritime surveillance, weather monitoring and other offshore applications.
12
59
3,057
The idea itself makes sense to me: most AI agents today ask us to describe the task and then hope the output matches what we meant. Sol flips that, it goes and finds the task in the inbox, does the work and waits for a yes before anything goes out.Â
What I want to see is how well it handles ambiguity. How will it perform outside a controlled demo?
The obvious is missing.
So we built Sol - hellosol.app
Sol finds the work itself, does it, and comes back for your approval.
Every day in our emails we say "I’ll share”, "I'll review”, "I'll get back" - then repeat the exact same thing to an AI. Why?
Sol finds everything you said you’d do & gets them started for you.
It does the research, creates the doc, builds the slides, finds the time, connects the dots across multiple emails, doing everything it takes to get the job done - but doesn’t send, schedule, or share anything until you approve.
Sol runs on its own computer, uses a browser, and has a library of skills that automatically get assigned to the work that needs to get done. No setup. It just starts working.
We've raised $4M from General Catalyst, Nexus Venture Partners, DeVC, PeerCheque, Kunal Shah, and a few others.
Extending early access now.
@generalcatalyst @nexusvp @DeVC_Global @peercheque
@neerajarora @b_jishnu @kunalb11 @miten @RTinkslinger @Rahul_J_Mathur @AkarshS27 @SiddhantD06 @RajatAgarwal167
1
1
7
1,426
Humanoid robots can now learn skills like climbing, balancing, and handling objects.
MotionDisco lets them learn without human demonstrations or remote control.
AI and motion planning help robots discover new ways to complete tasks.
9
41
207
13,757
Verobotics has developed a robot that can climb skyscraper walls to inspect and clean buildings.
Using two robotic legs and AI, it can inspect building exteriors up to four times faster than humans.
The collected data can also help predict maintenance needs while keeping people away from dangerous heights.
5
18
72
3,822