Neo humanoid maker 1X releases world models to help bots learn what they see

1X, the robot company behind it Neo humanoid robothas unveiled a new AI model that it says understands real-world dynamics and can help bots learn new information on their own.

This physics-based model, called 1X World Modeluse a combination of video and prompts to give the Neo robot new abilities. The video allows the Neo robot to learn new tasks it has not been trained in before, according to 1X.

This release comes as 1X gets ready to release the Neo humanoids into the home. The company opened pre-order for its humanoid in October with plans to ship the bot within the year. A 1X spokesperson declined to share a timeline for when the bots will ship or share information about how many pre-orders have exceeded expectations.

“After years of developing the world model and making Neo’s design as close to human as possible, Neo can now learn from internet-scale video and apply that knowledge directly to the physical world,” said Bernt Børnich, founder and CEO of 1X in a statement. “With the ability to turn directions into new actions – even without previous examples – this is the starting point of Neo’s ability to teach itself to master almost anything you can think of.”

Say that bots can edit any directions to a new act of high claims and not entirely accurate; you can not tell Neo to drive the car and will suddenly know how to park parallel, for example. But there is some learning.

1X did not say whether the world model allows the current Neo bot to perform new tasks directly from video capture and prompting, a company spokesperson said. Instead, the bot captures video data linked to specific directions and sends it back to the world model. The model is then fed back to the bot network to give it a better understanding of the physical world and better understand it.

It also gives the user insight into how Neo thinks about an action or reacts to certain prompts. That kind of behavioral information can help 1X train the model to the point where the robot will be able to respond to instructions in ways it has never done before.