Pre CES 2026 Humanoid Robot Tech Upends AI Industry (AI NEWS)

And they’re saying that this will ensure safety and effectiveness and they’re also trying to use this as a supervision and feedback loop to allow these robots to improve and adapt over time and then deploy them with full autonomy in the future. But tell us in the comments below whether or not you would allow for a human operator to be present in your house with you potentially 24 hours a day. Next we also have this robot from Persana AI and they’re calling it the Gen 1 humanoid robot. So here’s a sneak peek not only the robot but also their tendon-driven hand.

So as some background in 2025 they’ve already raised 42 million dollars in pre-seed funding and they’ve signed with HD Hyundai subsidiaries for the development and deployment of shipyard welding tasks in humanoid robots. And this will rely heavily on the success of their five-fingered dexterous hand that they gave a sneak peek of in their video. But here’s what looks to be the most impressive of all. This is Ubtex S2 humanoid and it’s using a dual-loop AI system here to play tennis and it pairs with the brain net 2.0 system with a co-agent architecture which allows for this robot to use a type of swarm intelligence in real-world applications.

So you could potentially have multiple robots playing tennis with each other in theory. But what does this mean in practice? Well the system replicates human binocular stereo vision and it gives the Walker S2 depth perception as well as environmental awareness which is comparable to that of a human. And on top of this the robot also comes equipped with generation four industrial dexterous hands which feature 11 degrees of freedom and a six tactile sensor array which allows each robot hand to lift up to 7.5 kilograms or over 16 pounds while individual fingers can handle objects up to one kilogram in weight or just over two pounds.

But not to be outdone by Ubtex S2 enter the playing badminton. And something to note here is that this demonstration is fully autonomous with no teleoperation. So this robot is able to autonomously predict where the shuttle cock will be and return real-time responses move through the space like a human with stable footwork and swings and return the ball. And the racket speed here is about 12 miles per hour or about 5.3 meters per second with an outgoing shuttle speed of about 43 miles per hour which so far is the fastest return speed of any robot that I’m aware of.

But share down in the comments if you know of any other badminton demonstrations that have gone beyond this. And in terms of reaction time this robot is reacting in sub-second speed and it’s capable of rallying with humans and demonstrating return shots in these initial demos that were shown with stationary positioning. And Neuralink just crossed a major threshold taking its brain computer interface technology beyond the US borders for the first time ever. This is because the company performed its first ever international procedure at Toronto’s University Health Network in Canada implanting devices in two patients with cervical spine cord injuries.

But the expansion doesn’t stop there because Neuralink also launched clinical trials in the United Kingdom through partnerships with University College London hospitals and Newcastle hospitals with the first UK participant known only as Paul becoming an example of how quickly the technology can take effect as he was controlling a computer using only his thoughts just hours after surgery. And beyond just computer control Neuralink also began testing more ambitious applications through its convoy project where participants started learning how to control robotic arms using their brain computer interfaces moving the tech from screens and cursors into physical manipulation.

And while still in its early phases this work hints at a future where people with paralysis could regain the ability to interact with the physical world grasping objects performing daily tasks and reclaiming independence. But Neuralink also unveiled an upgraded surgical robot with notable improvements where the new system inserts electrode thread in just 1.5 seconds and can reach depth succeeding 50 millimeters. Plus engineers designed it to be compatible with 99% of anatomical variations globally meaning it could work for nearly anyone. And luckily manufacturing costs for the needle cartridges have dropped by 95% which is a change that could prove critical as the company scales up.

So moving forward into 2026 Neuralink states that its goal is to make its routine as LASIK meaning just potentially minutes long using flexible electrode threads that are thinner than a human hair. And here’s something that’s genuinely impressive. This is Claude Opus 4.5 creating 3D models from scratch all by itself. And it’s not using vision here. It’s actually working blind using a custom set of tools or functions that it created for itself so that it can read a text based list of everything in the scene and understand the context that it’s working with.

So it actually built its own bridge using an MCP and used a web socket connection to link to its code editor where there it was able to do all of this cool work. So if you were to hook this up to a 3D printer imagine what could happen here. The last here I’m going to a leading South Korean humanoid robotics company and here are two clips. The first of which is this robot running with shoes on outside in somewhat uneven terrain as you can see here. And finally this robot is running on a treadmill.

And while we don’t know exactly how fast the robot is going or how long it can run, it is quite interesting to see that they’ve made this much progress so far and the way that it moves looks promising. But comment down below if you know any more stats around this robot. Like and subscribe and check out this video here for more on the latest in AI news. [tr:trw].