1X’s Neo Humanoid Robots Gain Self-Learning Abilities with New AI Model
Norwegian robotics company 1X has introduced the 1X World Model, an advanced AI system designed to enhance the learning capabilities of its Neo humanoid robots. This physics-based model enables Neo to interpret and learn from video data, allowing the robots to acquire new skills autonomously.
The 1X World Model integrates video inputs with specific prompts, facilitating Neo’s ability to perform tasks beyond its initial programming. By analyzing video data associated with particular commands, Neo can develop a deeper understanding of various tasks, thereby expanding its functional repertoire.
This development aligns with 1X’s strategy to deploy Neo humanoids in residential settings. The company began accepting preorders in October, with plans to commence shipments within the year. While specific delivery timelines and preorder figures remain undisclosed, a company spokesperson indicated that the response has surpassed expectations.
Bernt Børnich, founder and CEO of 1X, emphasized the significance of this advancement:
After years of developing our 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. With the ability to transform any prompt into new actions — even without prior examples — this marks the starting point of Neo’s ability to teach itself to master nearly anything you could think to ask.
While Neo’s capacity to translate prompts into actions represents a significant leap, it is important to note that the robots are not yet capable of executing complex tasks, such as driving a car, solely based on video input and prompts. The current system allows Neo to process video data linked to specific commands, which is then incorporated into the world model. This iterative process enhances the collective understanding and capabilities of the Neo robot network.
Additionally, the 1X World Model provides insights into Neo’s decision-making processes, offering users a clearer perspective on how the robot plans to respond to various prompts. This transparency is crucial for refining the model, with the ultimate goal of enabling Neo to handle tasks it has not previously encountered.
The introduction of the 1X World Model signifies a pivotal moment in the evolution of humanoid robotics, highlighting the potential for robots to learn and adapt in real-world environments through advanced AI integration.
