The race to build a functional, affordable humanoid robot is heating up. 1X, a California-based company, believes its Neo robot – priced at $20,000 – is a significant step forward. Unlike many robotic projects focused on industrial tasks, Neo is designed as a general-purpose home assistant capable of learning and adapting to real-world environments. The company’s CEO, Bernt Børnich, says the key is not pre-programming every action but giving the robot the ability to figure things out on its own.
The Core Concept: A “World Model” for Robots
1X’s approach centers around a “World Model” – a computer program that simulates the robot’s surroundings. This isn’t just about recognizing objects; it’s about predicting how they behave. For example, Neo doesn’t just see a Post-it note; it understands that paper can be peeled off a surface and read.
This is crucial because the real world isn’t a clean laboratory. Most robots struggle with unexpected situations, but 1X believes that by training Neo on vast amounts of human behavior data, it can mimic human reasoning. The robot learns from roughly a million hours of video footage showing people doing everyday tasks, including messy, real-world interactions like opening cabinets or handling fragile objects.
How Neo Learns: Imitating Human Interaction
Neo is physically designed to be similar to a human, which 1X argues is no accident. The premise is that the world is built for humans, and by mimicking human form and movement, the robot can leverage existing knowledge about how things work. It doesn’t just grab objects; it adjusts its grip carefully, like a person would.
This approach relies on the idea that a robot doesn’t need to be explicitly taught every task. Instead, it can generalize from what it observes. The company’s early demos showed Neo struggling with autonomy, relying heavily on remote human control. But recent tests demonstrate the robot performing tasks like toasting bread, watering plants, and even plunging toilets without direct intervention.
Limitations and Trade-offs
Despite the progress, Neo isn’t perfect. Tasks take minutes instead of seconds, and the robot still occasionally falters. Long-term memory – the ability to recall past conversations or preferences – remains a significant challenge.
1X acknowledges these limitations. Customers who opt out of data sharing (to protect privacy) may experience reduced capabilities, as the system relies on user data to improve. When Neo encounters unsolvable situations, a human operator can remotely guide it through a virtual-reality headset. While this adds cost, 1X believes teleoperation will become less necessary as autonomy improves.
The Future of Humanoid Robots
The push for humanoid robots isn’t unique to 1X. Companies like Tesla, Boston Dynamics, and others are pursuing similar goals. The trend at CES this year highlighted the growing momentum, with LG’s AI-enabled robot, Boston Dynamics’ Atlas, and other advanced models stealing the show.
The key question is whether these robots will become truly useful outside controlled environments. 1X is betting that its World Model approach will deliver on that promise, but it’s a long-term project. The first 10,000 customers who reserved a Neo for $200 are essentially early adopters willing to participate in the development process.
Ultimately, the success of Neo – and the broader humanoid robot revolution – hinges on whether these machines can reliably integrate into everyday life without requiring constant human intervention.
