Brett Adcock, founder and CEO of Figure - an AI Robotics company building the world's first commercially viable autonomous humanoid robot has announced the launch of a groundbreaking machine learning model designed for humanoid robots. The newly introduced model, called Helix, is a cutting-edge Vision-Language-Action (VLA) system that enables robots to understand and execute commands based on visual and language cues.

What is a Vision-Language-Action Model?

VLAs are an emerging technology in robotics, combining vision and language comprehension to enhance machine learning capabilities. The best-known example in this field so far is Google DeepMind’s RT-2, which trains robots using video data and large language models (LLMs). Helix operates on a similar principle, integrating visual input and language prompts to enable real-time robotic control.

Meet Helix, our in-house AI that reasons like a human

Robotics won't get to the home without a step change in capabilities

Our robots can now handle virtually any household item: pic.twitter.com/Wsx5s8Qelc

— Figure (@Figure_robot) February 20, 2025

Humanoid Robot Helix: How it Works

Helix is designed to bridge the gap between language processing and vision, allowing users to give robots simple verbal commands that they can instantly interpret and execute. According to Figure, Helix demonstrates remarkable object generalisation, meaning it can identify and handle thousands of household objects with varying shapes, colours, sizes, and material compositions—without prior exposure during training. Imagine telling a robot, “Pick up the red mug from the table,” and watching it perform the task flawlessly. Helix makes this level of seamless interaction possible by enabling the robot to visually assess its surroundings, understand commands, and execute tasks accordingly.

Helix Robots, One Unified System

One of Helix’s standout features is its ability to coordinate two robots simultaneously. Figure demonstrates this capability with examples such as “Hand the bag of cookies to the robot on your right” or “Receive the bag of cookies from the robot on your left and place it in the open drawer.” These interactions showcase how multiple robots can work in unison to complete complex tasks.

Helix as a Housemaid Robot

Figure is currently testing Helix with its 02 humanoid robots in a home environment. Unlike structured factory or warehouse settings, households present a dynamic and unpredictable landscape, making robot adaptation significantly more challenging. One of the biggest hurdles in household robotics is teaching machines to navigate and interact with ever-changing surroundings while maintaining reliability. Additionally, the high cost—often ranging from five to six figures—has made home robotics a lower priority for most companies. The prevailing strategy in the industry is to refine robots for industrial applications first, improving efficiency and reducing costs before introducing them into homes.

The Future of Home Robotics

While humanoid robots aren’t yet commonplace in households, advancements like Helix indicate that day isn’t too far off. For now, Figure continues to push the boundaries of AI-driven robotics, with a strong focus on industrial applications before eventually bringing its technology into everyday homes. With Helix, Figure is not just advancing robotics—it’s laying the groundwork for a future where machines seamlessly understand and execute human instructions with natural ease.