Unitree Rival Fourier Unveils N1 Humanoid Robot With 23 DoF (NEW AI ESPORTS TECH)

Because unlike most robots, For Your’s N1 is offering an open source approach, meaning that they’re not just selling the robot, but instead they’re also sharing the entire recipe to build it from the ground up, including a bill of materials that lists every part you need to make, as well as all the drawings for its manufacturing, and a step-by-step assembly guide, plus video tutorials to guide you all the way through. And it also comes with the software you need, which is hosted on GitHub with communication interfaces to plug your project into.

And on top of this, For Your is also planning to offer inference code and training frameworks in the near future. But what are the robot specs entering part two, the N1’s hardware? So the N1’s body, starting with the size, it stands at 1.3 meters tall, which is about 4 feet 3 inches, and it weighs 38 kilograms or roughly 84 pounds. And for your reference, this is slightly taller than Unitree’s G1, standing at just 1.2 meters or 4 feet tall, meaning that the N1 is probably small enough to fit in tight spaces like the trunk of your car, or even possibly inside of your luggage when traveling on a plane.

As for its structure, the robot’s made of aluminum alloy, as well as some plastic composite, which For Your says gives the robot extra strength without weighing it down, plus some extra flexibility for smoother movement. And as for the robot’s joints, it comes with 23 degrees of freedom throughout, meaning that it moves with 23 separate joints to let it twist and contort itself into positions that probably are beyond what even a human could achieve. But what kind of muscle does the N1 have to tie this all together, entering part three, the N1’s power system.

So For Your actually highlights three areas in particular, with the first being its actuators. Specifically, For Your is using its own custom FSA 2.0 integrated actuators, where FSA stands for For Your Smart Actuator, which is their fancy way of saying that they’re using proprietary tech to achieve more precise control. And these actuators run on their own control system setup that keeps movements stable and responsive. And as for the robot’s batteries, the N1 features a plug-in battery design to fit into the robot’s back for over two hours of continuous runtime. And when it comes to actually physically running, the robot can achieve a max speed of 3.5 meters per second, or about 7.8 miles per hour, which is faster than most people jog.

And For Your says the robot is built for rugged use, meaning that it can be bumped and jolted around without falling over easily, allowing developers more freedom to test new algorithms or motions without encountering hardware hiccups or limitations. And now for the robot’s results, part four, its real world performance. And this is where the N1 seems to stand out against other robots, as For Your has put the N1 through the ringer to verify these stats, starting with its endurance, where it logged over 1000 hours of outdoor testing on complex terrain.

Plus the robot was made to run for 72 hours continuously without stopping. And as for mobility, the N1 can climb 15 to 20 degree slopes and tackle 20 centimeter stairs, which is about eight inches per step. And it can stand on one leg or else it can quickly stand up if it falls over. And when it comes to stability, the N1 features a control system for more precision and anti-interference to keep it steady, even when it’s working inside of chaotic environments. But now for the meat and potatoes part five, the N1s artificial intelligence.

Right now the N1s open source software covers its basic operations. But for your says it’s working towards more advanced AI integrations with the following releases planned on the horizon, starting with its inference code, where for your will release code for running different AI models for the N1 process data, like vision and speech. And they’re also going to introduce training frameworks for developers to build custom AI solutions for tasks like object recognition or complex motion planning. Plus they plan on multimodal integrations, meaning that the N1 will eventually be able to combine its senses of sight, sound and touch to make smarter decisions in real world scenarios.

And as for the robots price, while there’s still no word on it yet, the company is probably going to try to compete with unitries 16 to $70,000 price range, meaning that this will probably be one of the more affordable robots for hobbyists, developers and businesses. But robots are usually only as good as their hands are dexterous, which brings us to the newest x hand esport hand from robot era. And it has several new features pushing the boundaries of what’s possible starting with the robot speed of 10 clicks per second, which is around that of a pro gamer.

And it’s probably way faster than you can click the mouse yourself, but comment down below if you can get anywhere close. Plus the hand also features full direct drive with no under actuation, meaning each actuator is actually directly connected to the joint that it controls without gears, belts or other transmission mechanisms. And in terms of its strength, they can handle a 24 kilogram payload or about 53 pounds. Plus it features compliant force control to adjust its force and motion in response to external interactions. And it has ultra low damping back drive with only gravity, meaning its joints or actuators can be moved with minimal resistance.

But the robot sensitivity is next level featuring a 600 pixel high precision 3d force sensor array, and it encompasses 270 degrees of the fingertip. And it can all be controlled with an Apple Vision Pro. And what’s really cool, I think, are the added touch of these little green sensor lights that come in different parts of the things. And the robot even features dynamic hand manipulation, meaning that it can basically figure out the best way to hold something on the fly that it hasn’t necessarily encountered before. And the hand is designed to pop on to other robots, potentially including sanctuary’s phoenix, as they’ve also just updated their robots touch sensors in its hands to provide a whole new level of dexterity.

And sanctuary AI is using Nvidia’s Isaac lab with Isaac sim to train these hands. And they’re powered by physics. And one of their standout achievements is the fact that their reinforcement learning train hand successfully performed a 500 gram object reorientation against gravity, which marks the first time that a five fingered robot hand has achieved this from simulation to real world transfer. And this is because of the low dampening back drive mechanics. And this hand is using miniaturized hydraulic valves to give it superior strength, speed and control over many other robotic hands.

And finally, opening AI just announced the launch of its GPT 4.1 4.1 mini and 4.1 nano, which are available only via API for developers not in chat GPT, and they outperform GPT for Oh, with 54.6% on software engineer bench verified, and this is versus a 33.2% score for GPT for Oh, and they also feature better multi step reasoning. Plus all these models support 1 million tokens instead of the previous 128,000 token limit, and GPT 4.1 mini is 83% cheaper, quite as fast as GPT for Oh, and nano is optimized for low latency tasks like auto complete.

And as for pricing, GPT 4.1 cost $2 per million input tokens and $8 per million output tokens, which is 26% cheaper than GPT for Oh, while the nano model costs 10 cents per million input tokens with 75% caching discounts. Anyways, like and subscribe and click this video here if you want to see more of the latest in AI news. [tr:trw].