Two weeks after splitting from Openai, the Humanoid Robot Maker figures have announced their latest breakthrough. This is an AI system that allows robots to understand speech and process objects they have never encountered before.
On Thursday, the diagram introduced Helix, a vision language action (VLA) model that allows humanoid robots to respond to voice commands and interact with unfamiliar items. The move shows numerical confidence in internal technology after Openai’s disbandment.
The helix runs on a 7 billion parameter model for understanding and 80 million parameter model for mobility control. In a recent demonstration, two robots worked together to organize groceries that they had never seen before, following natural language instructions. Unlike previous AI systems, Helix runs efficiently on a standard onboard GPU, requiring only a small portion of what previous models need.
“We present Helix, a generalist vision language action (VLA) model that unifies perception, language understanding, and learning control to overcome multiple long-standing challenges in robotics,” says Dialog AI. I mentioned it in the post.
What is the difference between Helix? Breakthroughs in AI-driven humanoid robotics
Helix addresses long-standing challenges in robotics by combining perception, language processing and control into a single unified system. The main features are:
Full Epper Body Control – Real-time movement of the torso, head, wrists and fingers allow for more precise operation. Multi-Robot Collaboration – Two robots can work together on shared tasks without pre-programmed adjustments. Generalized Object Processing – Helix-equipped robots can use voice commands to understand and manipulate objects they have never encountered. Unified Learning Model – A single neural network handles all tasks that work in sync with object interactions, open appliances, and other robots without undergoing task-specific fine-tuning. Deployment-Ready – Runs on a low-power embedded GPU, making it suitable for real applications.
Why is it important?
Industrial robots already perform accurate repetitive tasks, but adapting them to your home remains a challenge. Unlike a controlled industrial environment, objects that are unpredictable in a home (glassware, crumpled laundry, scattered toys) require robots to adjust on the spot.
Traditional training methods require demos or manual programming for thousands of people, making widespread adoption unrealistic. Helix changes this by dramatically reducing the time and effort required to teach robots to new behaviors, bringing home robotics closer to reality.
Just a year ago, Sunnyvale, California-based AI raised $675 million in funding in pre-money valuation, which rose significantly from its $400 million valuation in May 2023. It is supported by investors such as Jeff Bezos and Nvidia. The substantial funding reflects the growing interest in general-purpose humanoid robots and the potential to rebuild a variety of industries.
Source link
