The maker of the Neo humanoid 1X is releasing a universal model to help robots learn what they see

1X, the robotics company behind New humanoid robotIt has unveiled a new model of artificial intelligence that it says understands real-world dynamics and can help robots learn new information on their own.

This physics-based model is called 1X Universal Modeluses a combination of videos and prompts to give Neo robots new abilities. Video allows Neo robots to learn new tasks they have never been trained for before, according to 1X.

This release comes as 1X prepares to launch its new in-home humanoids. I opened the company Pre-orders on humanoids in October with plans to ship the robots this year. A 1X spokesperson declined to share a timeline for when these robots will ship or share any information regarding how many have been ordered other than to say that pre-orders have exceeded expectations.

“After years of developing our global model and making NEO’s design as human-like as possible, NEO can now learn from internet-scale video and apply that knowledge directly to the physical world,” Bernt Bornisch, founder and CEO of 1X, said in a company statement. “With the ability to turn any prompt into new actions – even without prior examples – this represents the starting point for Neo’s ability to teach himself to master almost anything he can think of.”

Saying that the robot can transform Any prompt In a new measure it is a noble claim and not entirely accurate; You can’t ask Neo to drive a car and he’ll suddenly know how to parallel park, for example. But there is some learning happening.

A company spokesperson clarified that 1X is not saying that the universal model allows existing Neo robots to immediately undertake a new task from capturing video to prompting. Instead, the bot takes video data associated with specific prompts and sends it back to the global model. This model is then fed back into the botnet to give them a better understanding of the physical world and more knowledge.

It also gives users insight into how Neo might think to act or respond to a particular prompt. This kind of behavioral information can help 1X train these models to the point where the bots will be able to respond to a prompt for something they haven’t done before.