US-based robotics company Figure announced that their humanoid robot, capable of assisting with household chores, has reached a level where it can be controlled through voice commands thanks to a new machine learning model. Brett Adcock, the founder and CEO of Figure, explained that their Visual-Language-Action (VLA) machine learning model, named Helix, enables robots to combine visual and language commands to provide real-time responses. Helix’s ability to recognize objects in the home and interact with them allows it to even use items it has not encountered before. This model, which can control two robots simultaneously, has the potential to overcome many challenges faced in assisting humanoid robot technology for household chores.

How does it work?
Helix combines visual data and language requests to control a robot in real-time, bridging the gap between vision and language processing. When a robot receives a voice command through Helix, it evaluates its surroundings visually and then carries out the task. Figure has demonstrated through video examples that their humanoid robot 02 can understand and execute voice commands like “Give the cookie bag to the robot on your right” or “Take the bag from the robot on your left and put it in the drawer.”

Robots empowered by the Helix machine learning model can also distinguish specific objects requested through voice commands from a variety of household items, and pick them up from among items with its hands.

A Challenging Process
One of the key reasons delaying the use of humanoid robots in homes is the difficulty in learning and controlling them. For humanoid robots to be useful in the home, they need to be able to generate new behaviors for objects they have never encountered before. Adcock, the CEO of Figure, emphasizes the need for manual programming at a doctoral level by many individuals for even a single behavior to be taught to robots. Therefore, manual programming in environments with many variables like homes is not feasible. Training robots with machine learning models stands out as an alternative. Robots can be trained with hundreds of hours and tens of thousands of repetitions required for learning.

Experts point out that robots need to have performed an action hundreds of times in the past to be able to do it on the first try.

The Biggest Barrier: Cost
Serious technological advancements are being observed in the development of humanoid robots for doing household chores, which have been a major dream of robotics companies and science fiction enthusiasts. Figure aims to popularize the use of home robots with the help of their innovative technology Helix.

According to experts, the main barrier to the widespread adoption of home robots that could simplify daily life is the hefty price tags that can run into six figures. Companies in the sector are all looking for ways to reduce costs.

It is highlighted that for humanoid robots, which are starting to become widespread in industrial settings, to enter homes and be demanded by large masses, they need to become more appealing in terms of price/performance ratio.