Phantom MK1 vs EngineAI T800 Humanoid Robot Warriors (AI NEWS)

And just for reference, a human can deliver between two to 400 newton meters, but probably closer to 300 for an athlete. So would this be able to defeat a human if in the ring, so far they’re delivering some of these kicks and punches with a good amount of agility? And looks like they’re going for round two here. And just for some reference here, both of these robots feature AGX Orin with 275 tera operations per second and a 360 degree lidar system, as well as multiple depth cameras to guide them. Is this real? Is this all real footage? Or is some of it CGI? Comment down below.

And we see these robots trading blows here. At some points, pieces flying off. Starting price of the T-800 is $25,000. So is this going to become a new sport? And will betting on the outcomes of robot fights become more popular than those of humans? But what if you give one of these robots a weapon? That’s what robot era just did, giving its L7 a sword. And it’s doing this demonstration by itself. In just a moment, a human will join it and you can compare both. This robot stands at 171 centimeters and features 55 degrees of freedom throughout its body.

But look at the way that it can jump through the air. It’s pretty impressive. 400 newton meters of torque. And of course, its joints, you can see they’re rotating 360 degrees with 20 kilograms of dual arm payload. And it can run at four meters per second. And looking at this robot perform, one thing to note is that it’s not a real sword. How would this be able to hold up with the extra weight? All together, looks like it’s able to handle itself pretty well. The hands here are 12 degrees of freedom, and they actually feature up to 10 clicks per second of speed.

So these are ultra dexterous, which would probably help in these domains of using different weapons. And speaking of weapons, the US military has deployed its first humanoid robot. Thanks to Foundation Robotics. This is the Phantom MK1. And this robot stands at 1.75 meters tall, weighs 80 kilograms, and it has a walking speed of 1.7 meters per second, or about six kilometers per hour. And right now, it’s already helping the Air Force, Navy, and Army for tasks like aircraft refueling, maintenance, and breaching operations. And Foundation has also shown a preview of this new hand that will be integrated into the Phantom MK1.

And they’re going to do a deep dive, but this is designed for handling weapons, as well as doing heavy duty tasks and industrial setting. And so far, it seems like it is tendon driven. But there aren’t too many specs on this hand as of yet. So comment down below if you have more information. Is a military robot a good or bad idea long term for humans? And because controlling these robots that may or may not be armed in the near future is so important, tele AI just released TextOp, which is a breakthrough that allows humans to control these humanoids with speech alone.

So put simply, TextOp is a real time text based control system for humanoids. So instead of using complex programming or awkward controllers, humans can simply use verbal instruction, and the robot will execute it seamlessly, shifting through moves like as shown in this demo, and TextOp employs a dual layer design. The first is a high level component that translates the text in a dynamic motion trajectories instantly. And the second is a low level controller that manages precise, full body stability on the actual robot. And the real innovation here is its streaming inputs, which lets users revise directions in the middle of an action with the robot adapting fluidly between tasks.

But control may come down to dexterity, which is why Hungarian robotic startup Alonic is drawing attention with their new AI hardware called the Alonic Hand. And this is a soft bio inspired robotic manipulator that’s built using new manufacturing process that’s called 3D tissue braiding. So rather than assembling the robot from hundreds of rigid parts, the company is weaving fibers, tendons and structural elements into a single integrated body through an automated production process. And this results in a compliant robotic hand that’s both strong and safe for human interaction. And this tech could solve a manufacturing bottleneck in robotics where Alonic’s automated platform could produce these complex robotic mechanisms in minutes instead of weeks, which would potentially lower the costs and increase the overall production of the robotics.

And while it’s still in its early stages, they’ve already raised $7.2 million to continue development and further expand on pilot deployments. And finally, in generative AI, making movies may now be as easy as typing out a simple idea as ByteDance just released its Seed Dance 2.0 in beta, which just took generative AI video to a whole new level much sooner than expected. And that’s because Seed Dance 2.0’s generation system now supports up to four types of input simultaneously. That’s images, videos, audio and text. In fact, creators can even mix as many as nine images, three videos and three audio clips together for a maximum of 12 files overall, with output videos lasting between four to 15 seconds each, including built-in sound effects or music by default.

And something to note is that while these demo clips were provided by ByteDance, they were undoubtedly selected from a bigger set of outputs, so details on everyday consistency, pricing and generation times remain unknown for now. But ByteDance says that Seed Dance 2.0’s key innovation is in its reference features, which allow the model to extract camera angles, motions, as well as effects from provided reference videos, as well as substitute characters and even extend sequences. Furthermore, ByteDance supports editing actions such as inserting or swapping out characters, but due to regulatory requirements, realistic human faces are currently restricted in input uploads at present.

Anyways, you can check out Seed Dance 2.0 right now in beta, but like and subscribe and check out this video here for more of the latest in AI news. [tr:trw].