Helix Tidies the Living Room
A little over a month ago we released Helix 02, a single neural system that controls the full body directly from pixels, enabling dexterous, long horizon autonomy across an entire room. After cleaning up a kitchen, Helix 02 is now taking on another everyday task: tidying up a living room.
If you could give a home robot one job, “tidy the living room” would be near the top of the list. But from a robotics perspective, this task is incredibly difficult. Unlike more structured commercial tasks, a living room changes constantly. Objects are scattered unpredictably. Furniture creates narrow navigation paths. Soft items like towels and pillows behave dynamically. Many actions require both hands, while others require freeing a hand in the middle of a task. And nearly every behavior involves moving through the room while manipulating something at the same time.
In this new demonstration, Helix 02 performs whole body, end-to-end living room cleanup - walking through the room while continuously manipulating objects, tools, and containers.
Key Results
Helix 02 continues to learn new tasks that demand the full integration of locomotion, dexterity, and sensing just by adding new data. With no new algorithms, no special-case engineering Helix learned to:
Clean surfaces with coordinated tool use: Use a spray bottle to wet a dirty surface, then perform forceful wiping motions with a towel to remove the mess.
Handle flexible objects dynamically: Manage the complex dynamics of a towel - unhooking it from the arm, repositioning it for cleaning, and whipping it over the shoulder to free its hands.
Perform complex bimanual manipulation: Pick up a bin with both hands and hold it while scooping blocks from a table into the container.
Use whole-body strategies for efficiency: Tuck a container under one arm to free both hands for picking up toys.
Execute dynamic object throws: Toss a pillow back onto a couch with a fast, controlled motion.
Perform in-hand reorientation for precise tasks: Pick up a remote, reorient it in-hand, and press the correct button to turn off the TV.
Reorganize tools during motion: Temporarily stow a towel under an arm while transitioning between tasks.
Navigate tight spaces with precise foot placement: Side-step through the narrow gap between a coffee table and couch while continuing manipulation.
Why This Matters
Tidying a living room combines locomotion, dexterous manipulation, tool use, and real-time planning in a constantly changing environment.
Helix handles all of these behaviors with the same general-purpose architecture used for previous tasks. Rather than engineering specialized controllers for each behavior, the system learns the strategies directly from data.
As more tasks are added, Helix continues to expand its repertoire—building toward a future where a single humanoid system can perform the wide range of everyday work required in homes and workplaces.
This is another step toward scalable humanoid intelligence: a single model that learns new capabilities simply by seeing more examples of the world.