Humanoid robots surpass 24-hour continuous autonomous work
Figure AI has proven that its vision of humanoid robots in the workplace is no longer just a theory. The three robots, affectionately nicknamed Bob, Frank and Gary by viewers of the live webcast, exceeded all expectations in a test. The original goal was an eight-hour work cycle, but due to perfect performance and no errors, the test was extended to a multi-day continuous operation that exceeded the 24-hour threshold.
The company's founder and CEO, Brett Adcock, emphasized that this is a historic achievement. During this time, the robots independently sorted more than 28 thousand packages, achieving speeds comparable to human workers. While a person spends an average of about 3 seconds on one package, the new F.03 model is right behind him. The entire process is based on visual perception; the robots read barcodes directly from the camera pixels, pick up the packages and place them correctly on the conveyor belt.
The key to this success is Helix-02, an internal neural network that runs entirely on the robot’s hardware. The company says there is no teleoperation or remote control involved – every move and decision is the result of artificial intelligence. The system also includes advanced mechanisms for autonomous problem solving. If the robot gets stuck or encounters an unknown situation, Helix-02 triggers an automatic reset, allowing the robot to continue working without human intervention.
Even more impressive is the self-maintenance capability of the entire fleet. If one of the robots detects a software or hardware error, it independently leaves the work area and goes for service, and its place is immediately taken by another robot, ensuring a smooth workflow. Figure AI thereby strengthens its role in the competition with companies such as Tesla and Agility Robotics, and builds on previous experience gained from testing at BMW plants. Helix-02 combines vision, touch and full-body control into a single model, which allows humanoids to achieve exceptional coordination in dynamic environments.






















