Living Room Ready: Watch How Figure 03 Tackles Domestic Tidying with Helix 02

In the rapidly escalating race to bring humanoid robots into the domestic sphere, Figure has released a new demonstration showing its Figure 03 robot autonomously tidying a living room. The 2:28-minute video, released today, highlights the robot performing a series of complex, interlinked tasks—from spraying and wiping down tables to tossing pillows and organizing toys—marking what CEO Brett Adcock calls a "major milestone towards a robot in every home."

Beyond the Kitchen Sink

The demonstration follows the unveiling of Helix 02 earlier this year, where the company showcased the Figure 03 loading a dishwasher. While the previous demo focused on a relatively stationary task, the living room cleanup demands what Figure calls "loco-manipulation": the ability to move through an environment while simultaneously manipulating objects.

According to Figure, the living room represents a unique challenge for autonomous systems. Unlike the structured environments of the BMW Spartanburg pilot, home environments are unpredictable, featuring narrow navigation paths between furniture and dynamic, "soft" objects like towels and pillows that do not maintain a rigid shape.

Key Capabilities Demonstrated

The footage highlights several sophisticated behaviors that Figure claims were learned through data rather than hand-coded heuristics:

• Coordinated Tool Use: The robot uses a spray bottle to wet a surface before performing forceful wiping motions with a cloth.
• Bimanual Manipulation: Figure 03 is seen picking up a bin with both hands to catch toys being scooped off a coffee table.
• Dynamic Object Handling: In a display of athletic coordination, the robot tosses a pillow back onto a couch with a controlled motion and "whips" a cleaning towel over its shoulder to free its hands.
• Precise Dexterity: The robot picks up a television remote, reorients it within its 7th-gen hands, and presses the power button.
• Whole-Body Efficiency: To maximize utility, the robot is shown tucking containers under its arm to keep its hands free for secondary tasks, a behavior often seen in human "multitasking."

The "Software 2.0" Strategy

The core of this demonstration is the Helix 02 AI architecture. Figure emphasizes that no new algorithms or special-case engineering were required to teach the robot these specific domestic chores. Instead, the system operates as a single, end-to-end neural network that translates "pixels to torque," allowing the robot to learn new tasks simply by being exposed to more demonstration data.

This approach is central to Figure’s goal of scaling to 100,000 units at its "BotQ" facility. By removing the need for 109,504 lines of C++ code—a feat Adcock announced in February—the company aims to create a "general-purpose" brain that can adapt to the "messy reality" of "unseen" homes.

The Road to 24/7 Autonomy

While the demo is visually impressive and remarkably smooth, it remains a "room-scale" success within a controlled environment. The ultimate test for the $39 billion startup will be whether this level of reliability holds up when the robots are shipped to customers.

Adcock has previously admitted to "babysitting" the robots around his own children, suggesting that while the "Software 2.0" transition is well underway, the safety and reasoning layers required for "free reign" in the home are still being refined. Nonetheless, the ability to side-step through narrow gaps while stowing tools and manipulating objects suggests that the gap between human and robotic domesticity is closing faster than anticipated.