Ready For Tomorrow #75

In partnership with

Robots are on the move – literally. One’s learning how to behave, another is delivering your groceries, and there’s even a hand that can throw a proper sign. Oh, and if you thought Lego was fun, wait until you see cubes that crawl under trees and build bridges. Buckle up, this week’s robotics news feels like a sneak peek into tomorrow.

Atlas learns behaviors, not just moves

Boston Dynamics and Toyota Research Institute just showed something really big. Their humanoid Atlas got a new AI system – Large Behavior Models.

Until now, robots, including humanoids, were like specialists in one thing. You program pick-and-place – they do pick-and-place. But what if something changes? Freeze. Humanoids were mostly for show – they could do a backflip or a dance, but in real work they didn’t hold up.

Now that’s changing. Atlas isn’t learning single moves anymore, but entire behaviors. A human in VR goggles shows it a task, and the robot then performs it on its own – and most importantly, it knows how to react when something goes wrong. If a part falls to the floor, Atlas picks it up and keeps working.

This is a breakthrough because, for the first time, a humanoid is starting to look like a universal worker, able to handle different tasks with one system, without the need for complicated code. It also opens the door to humanoids in factories and warehouses, not as a gimmick but as real support.

This is the moment when humanoids step out of the “flashy show” phase and into the “practical work” phase.

The World’s Most Wearable AI

Limitless is your new superpower - an AI-powered pendant that captures and remembers every conversation, insight, and idea you encounter throughout your day.

Built for tech leaders who need clarity without the clutter, Limitless automatically transcribes and summarizes meetings, identifies speakers, and delivers actionable notes right to your fingertips. It’s securely encrypted, incredibly intuitive, and endlessly efficient.

Order now and reclaim your mental bandwidth today.

Discover Limitless

Glovo couriers, watch out!

A new player has arrived in Austin, Texas, and it could shake up the delivery market. It’s Robomart – a company that has been experimenting with mobile shops on wheels for years, and now it’s rolling out seriously with its delivery robot RM5.

It looks like a small van, but instead of a driver it has 10 lockers for packages and drives straight to where the orders are. Each locker can hold food, groceries, or even larger parcels.

The most interesting part? Instead of commissions and tips like on UberEats or Glovo, Robomart follows one simple rule: a flat $3 delivery fee. That’s it.

This could be a breakthrough, because one robot can handle several customers on a single trip, which means cutting delivery costs by up to 70%. On top of that, there’s no need for an army of couriers on bikes or scooters – just a few of these electric vans per neighborhood.

And Robomart doesn’t just want to be another delivery company. Its goal is to build an entire network of mobile marketplaces – robots that bring your shopping right to your doorstep.

In short: Glovo couriers, watch out – the competition isn’t sleeping, it’s driving itself.

FieldAI and the “physics-first” brains for robots

The company FieldAI is introducing a new class of artificial intelligence for robotics – Field Foundation Models. These are “physics-first” models, designed from the ground up to account for the laws of physics and the real-world constraints robots face. Thanks to this approach, robots can operate without maps, GPS, or pre-set routes, handling dynamic and unpredictable environments on their own.

FFMs let a single “brain” control different types of robots – from quadrupeds to humanoids to wheeled vehicles. They can work with many types of sensors and are resistant to the hallucinations typical of language models, which makes them safer and more reliable. The real game-changer, though, is that they can generalize and adapt in real time, without retraining or manual coding of behaviors.

This is the first time AI has been built from scratch specifically for robotics, not as a tweak of existing language or vision models. FieldAI’s solutions are already being used in construction, energy, and transportation, and the company has secured more than $400 million in backing from investors like NVIDIA and Jeff Bezos’s fund.

Experts say FFMs could become for robotics what operating systems were for computers – a foundation that makes large-scale, safe deployment of robots in the real world possible.

A robotic hand that knows sign language

A new humanoid hand has been created – the Ambidextrous 23 Direct-Drive Humanoid Robotic Hand. The name alone sounds complex, but the technology is even more impressive.

This hand has 23 degrees of freedom, with each finger powered by its own servo, and the thumb equipped with five extra motions that give it human-like dexterity. What makes it even more unique is that it’s two-sided – with a single switch it can function as either a right or left hand, without any rebuilding or extra parts.

The most remarkable feature is its ability to perform the full alphabet of American Sign Language – and it can do this in both orientations. That means it could become a powerful educational and assistive tool for people learning or using sign language.

What matters here is that robotic hands are no longer just about gripping objects. With this level of precision and speed, they can also communicate with humans, opening the door to social and educational applications that reach far beyond industry.

In short: humanoids are gaining not only strong muscles but also increasingly capable hands – hands that can handle language, gestures, and finesse that robots have always lacked.

Modular robots like living building blocks

Researchers at Dartmouth College have developed innovative cube-shaped modular robots that can change shape and connect into larger structures. Each module has a 3D-printed core with a battery and Wi-Fi, eight rods, and flexible strings whose tension is controlled by a motor. This design allows the robot to deform, move, and latch onto others.

In tests, the modules formed ant-inspired bridges, scaffolding, and even improvised stretchers capable of carrying a mannequin. They were able to squeeze through narrow gaps, crawl under fallen trees, and even collaborate with a drone that assisted in repairs and building larger structures.

This nature-inspired solution has huge potential for rescue missions – the robots are lightweight, can be dropped from the air, and quickly assembled on-site into bridges, shelters, or infrastructure elements.

The modular robots from Dartmouth show that the future of robotics is not only about humanoids but also about self-configuring systems in the field, capable of adapting like living organisms.

That’s it for today’s roundup. In front of us there’s still a long road until the end of the week, so keep your heads up, take a breath, and let’s keep moving forward.

The robots will handle their part – now it’s our turn.

Cheers, Jacek!