The technological advancement of humanoid robots has reached a significant milestone. The company Figure AI presented Helix, a new vision-language-action (VLA) model platform for real-time control of humanoid robots via voice input. This innovation is characterized by the combination of visual and voice-based data processing as well as motion control and promises to fundamentally redefine the interaction between humans and machines.
Advanced technology – modular and efficient
At the heart of Helix is an innovative two-component system consisting of a multimodal language model with 7 billion parameters and a complementary motion AI with 80 million parameters. This implemented infrastructure enables humanoid robots to perform precise motion sequences while coordinating up to 35 degrees of freedom simultaneously. Particularly noteworthy is the ability to recognize and handle a variety of unknown household objects without having been previously trained on their specific characteristics.
Helix also sets new standards in terms of efficiency: with just 500 hours of training data, a system was created that performs tasks with impressive precision. Compared to other approaches, which often require many times more training time, this underlines the relevance and feasibility of Helix for commercial applications.
Potential applications in domestic environments
A key goal is the use of humanoid robots in private households to make everyday tasks easier, such as sorting groceries or collaborating on more complex tasks. A video impressively demonstrated this potential: two robots placed food in a refrigerator without specific pre-training data for the items used.
While many industry players are currently focusing on industrial or workplace-related robot applications, Figure AI is taking a strategically surprising approach with its household focus. By utilizing embedded-capable processing based on integrated GPUs, the company is developing Helix to be practical and market-ready for the broad consumer market.
Challenges and opportunities for the robotics industry
Despite this progress, there are still unanswered questions. The real-world performance of humanoid robots in uncontrolled environments remains a key challenge. There are also issues such as safety standards, product pricing and the clear definition of user applications.
The market for humanoid robots is growing dynamically: projections assume an average annual growth rate of 96% between 2022 and 2030. Helix not only brings technological progress, but could also make a significant contribution to promoting acceptance and integration by end users – provided that the aforementioned hurdles can be overcome.
The most important facts about the Helix innovation:
- Technology: Multimodal voice and activity system with 7 billion and 80 million parameters.
- Core capabilities: Real-time control of humanoid robots with 35 degrees of freedom of movement.
- Area of application: Optimization for household tasks through visual data processing and voice prompts.
- Efficiency: Short training time of only 500 hours with a wide range of applications.
- Commercialization: Integration on embedded GPUs for marketable products.
Source: Figure AI
