Figure trains its robot's stability controller entirely in a physics simulator, akin to a video game. This allows them to test countless scenarios synthetically. The resulting AI model is so effective it can be 'zero-shot' deployed directly onto the physical robot, achieving human-level stability immediately.

After realizing its initial tendon-driven hand design was an engineering dead end, the team pivoted quickly. Rather than wait months for a full redesign, they repurposed motors from the robot's feet to power the wrist, creating a 'Frankenstein' prototype that allowed AI development to continue without delay.

Unlike cloud-reliant AI, Figure's humanoids perform all computations onboard. This is a critical architectural choice to enable high-frequency (200Hz+) control loops for balance and manipulation, ensuring the robot remains fully functional and responsive without depending on Wi-Fi or 5G connectivity.

Instead of reacting to its environment, ONE X's world model AI allows its robots to 'think' forward and simulate potential outcomes of an action. Like a human anticipating spilling hot coffee, the robot can identify risks and select the safest trajectory, which is critical for operating in a home.

In robotics, purely imitating human actions is insufficient. A model trained this way doesn't learn how to recover from inevitable errors. Comma AI solves this by training its models in a simulator where they are forced to learn recovery paths from off-course situations, a critical step for real-world deployment.

Figure determined that coding robot movements is unscalable due to the infinite possible states (360^40). They pivoted from traditional C++ to Helix, an AI policy that controls the robot's entire body from camera inputs, treating robotics as a neural network problem, not a software engineering one.

Unlike fixed industrial robots, a simple emergency power-off is unsafe for humanoids. They require constant energy to balance, so an emergency stop would cause them to fall over, creating a new and unpredictable hazard. This fundamental difference requires an entirely new set of safety protocols for the industry.

According to Comma AI's CTO, the next frontier in robotics isn't just bigger models, but solving three fundamental challenges: 1) using ML for low-level controls, 2) making reinforcement learning (RL) practical for noisy environments, and 3) enabling continual, on-device learning to adapt to changing conditions.

Figure's robots do not rely on a cloud connection for their core functions. The Helix AI model runs inference on GPUs located inside the robot's torso. This allows them to perform complex tasks like logistics or tidying a house even if they lose network connectivity, ensuring high operational reliability.