Neo-humanoid creator 1X releases global model to help robots learn what they see

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

This physics-based model, called 1X global modeluses a combination of video and prompts to give Neo robots new abilities. Video allows Neo robots to learn new tasks they had not previously been trained to do, according to 1X.

This release comes as 1X prepares to release his Neo humanoids into the house. The company opened pre-orders for its humanoids in October and plans to ship the robots this year. A 1X spokesperson declined to share a shipping timeline for these robots or share information on the number of robots ordered, beyond saying that pre-orders were exceeding expectations.

“After years of developing our world model and making Neo’s design as close to human as possible, Neo can now learn from video at internet scale and apply that knowledge directly to the physical world,” said Bernt Børnich, founder and CEO of 1X in a company statement. “With the ability to turn any prompt into new actions – even without prior examples – this marks the starting point of Neo’s ability to learn to master almost anything you could ask.”

Say that the bot can transform any prompt embarking on a new action is a noble statement and not entirely accurate; you can’t tell a Neo to drive a car and it will suddenly know how to parallel park, for example. But learning is underway.

1X does not say that the global model allows today’s Neo robots to perform a new task immediately after capturing a video and after being prompted, a company spokesperson clarified. Instead, the bot takes video data related to specific prompts and then feeds it back into the world model. This model is then fed back into the network of robots to provide them with a better understanding of the physical world and more know-how.

It also gives users insight into how Neo plans to behave or react to a certain prompt. This type of behavioral information could help 1X train these models to a point where the robots will be able to respond to a prompt indicating something they’ve never done before.