In partnership with

Welcome back!

Another week, another batch of stories that make the robot world feel like a mix of sci-fi and a student project that accidentally went too far.

Grab a drink and let’s go!

TL; DR

• Lobster tail becomes a soft robotic gripper.

• Pudu D5 enters the stage as a true outdoor workhorse.

• EngineAI T800 looks like it skipped the polite humanoid phase.

• Yann LeCun leaves Meta to build world models and embodied AI.

• Humanoid robots can now run, but the market may be running without a plan.

Lobster 2.0 a natural part doing robotic work

Scientists at EPFL had an idea that many AGH students also get on a Thursday night at Club Studio, with one difference. They actually remembered it in the morning and did it. They took a lobster tail and turned it into a robotic gripper.

It sounds absurd, but the lobster tail has natural springiness that works perfectly for soft grippers. The researchers added thin wires and a simple control system, and suddenly the shell started acting like a delicate, flexible hand. In tests it can even pick up a mushroom. without crushing it into oblivion.

This is still an experiment, yet it feels like a preview of a new family of light and eco-friendly robot tools. Zero fancy materials. nature delivered a ready element, and the scientists just activated it. And suddenly a lobster tail becomes a solid robot part, not a souvenir from dinner.

Turn AI into Your Income Engine

Ready to transform artificial intelligence from a buzzword into your personal revenue generator?

HubSpot’s groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.

Inside you'll discover:

• A curated collection of 200+ profitable opportunities spanning content creation, e-commerce, gaming, and emerging digital markets—each vetted for real-world potential

• Step-by-step implementation guides designed for beginners, making AI accessible regardless of your technical background

• Cutting-edge strategies aligned with current market trends, ensuring your ventures stay ahead of the curve

Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.

Get Your Guide

Pudu D5. a quadruped that fears neither stairs nor rain

Pudu Robotics showed its new outdoor robots D5 at iREX 2025, and we finally have a machine that actually works outside a clean, flat hall. One version walks on legs, another drives on wheels, and both are meant for places that are wet, uneven, loud and uncomfortable.

Here industrial grade really means gear for harsh conditions. The robot operates in rain, dust, cold and heat. It has IP67 protection, handles 30 degree slopes and climbs stairs up to 25 cm. Where a typical service robot would fall over, this one just keeps going.

Inside there is a strong onboard computer that processes the environment in real time. In practice this means live terrain mapping, so the robot knows where it is, what is in front of it and which route is best. It uses LiDAR sensors that scan the environment with lasers and wide angle cameras that help it avoid missing anything important.

The robot moves up to 5 m per second, carries up to 30 kg and works around 2 to 3 hours on one charge. A perfect pick for inspections, patrols, logistics and all tasks where a human would rather not freeze or wade through mud.

In short. Pudu D5 sees no problem with rough terrain. it simply gets in and does the job.

EngineAI T800. the name says everything

EngineAI from Shenzhen presented its humanoid T800 and the name alone suggests it is not a robot built to serve coffee. This is a full sized machine designed for heavy work. lifting, inspection and logistics, meaning places where strength and stability matter most.

The biggest buzz came from the video released with the premiere. The T800 jumps into a kick, spins like in capoeira and throws punches with such confidence that the internet instantly called it CGI. EngineAI had to publish a second video. raw, without polishing. to prove the robot really moves like that.

There is more. T800 is scheduled to appear on December 24 in the Mecha King tournament, where its stability and dynamics will be tested in conditions similar to a ring. The company says these extreme trials translate into better industrial reliability later on.

T800 looks like a humanoid that does not ask where the kitchen is. it asks where the heaviest job needs to be done.

Yann LeCun leaves Meta. and the AI world heats up

It has been loud in artificial intelligence. The godfather of AI, Yann LeCun. pioneer of deep learning, creator of CNNs and Turing Award laureate. announced that after 12 years he is leaving Meta. Not quietly, but with a clear plan to start his own thing. LeCun is launching a Paris based startup focused on more advanced machine intelligence. the kind that does not just generate text like LLMs but understands the physical world.

LeCun has long argued that current language models are a dead end. they perform well on text but have no sense of causality or physics. This is why his new project will focus on world models. systems that can predict how reality works. A core component for future robotics and embodied AI.

He announced his departure after Meta restructured. Zuckerberg shifted the company to commercial LLMs and chatbots. Meta Superintelligence Labs was created, around 600 people were laid off from FAIR and long term research slowed down. LeCun thanked the company for the years together, praised FAIR for PyTorch and Llama, but made it clear he wants to build AI seriously. Meta stays a partner, but not an investor in his new firm.

His departure symbolically splits the field into two groups. those who believe in the future of LLMs and their commercial path, and those who think the real breakthrough will come only when machines learn to understand the world the way we do. In parallel there are fresh concerns about an AI bubble. weaker Llama 4, rising costs, talent leaving.

For startups this opens a new stage. especially in robotics and systems that learn the physical world. exactly the area where future humanoids will need real intelligence, not just polished answers.

The AI sector just got a signal. a new chapter begins, and once again LeCun wants to write the first line.

Humanoids started to run. the question is where

The past weeks have been a marathon in the humanoid world. EngineAI released its T800 that kicks, spins and looks ready for a holiday fight. At the same time Figure and Tesla showed that their humanoids can simply run. And not like robots from a few years ago that shook like a washing machine spinning full speed. they run smoothly.

Let us be honest. Figure won this comparison. Their robot moves so naturally you could believe it has been training Hyrox since forever. The step is natural, the rhythm stable and there is zero weird knee bouncing or chaotic arm flapping. Tesla made progress too, but Figure looks like an athlete and Tesla looks like someone at their first training.

One thing is clear. this world is speeding up. Robots run faster, companies run even faster and the hype runs the fastest. And that is where the concern starts. Because more voices now say that this might be a sprint straight into a wall. Even the Chinese government recently stated plainly. there are over 150 companies building humanoids because everyone else builds humanoids. No plan. no clear market. just inflating a balloon.

It is good that someone finally said enough. This market needs a sober look, because running alone is not enough. The real question is not who runs faster. it is who will actually deliver a useful robot. One thing is sure. the speed is cosmic, but if no one starts checking the path ahead, the finish line may turn out to be a wall.

That is it for this week.

Lobsters turned into tools, robots marching through storms, humanoids training for tournaments and AI pioneers starting rebellions in Paris.

The future keeps getting louder, stranger and somehow more interesting.

See you in the next episode.

Keep your sensors clean and your optimism charged.