Can embodied AI move from demos to real service use?

Question

Accepted Answer

Matrix Robotics says embodied AI can move from demos to real service use via its WAVE-powered humanoid coffee service at Summer Davos

Embodied AI can move beyond laboratory demonstrations into real, continuous service use through live, in-venue deployment, as shown by Matrix Robotics at the World Economic Forum’s 17th Annual Meeting of the New Champions in Dalian. The company, which participated as the sole humanoid robotics service provider, deployed its third-generation humanoid platform, MATRIX-3, to serve attendees as an AI barista during the June 23 opening of the “Innovating at Scale” forum. Its system delivered live coffee service using the WAVE Embodied AI Foundation Model alongside a general-purpose full-body motion control system.
The forum demonstration focused on real service, not static displays
Summer Davos ran under the theme “Innovating at Scale” and brought together more than 1,700 leaders from over 90 countries to discuss translating innovation into real-world applications. Matrix Robotics moved away from conventional static humanoid displays by building a live demonstration area in the venue’s first-floor dining zone. As the forum’s only humanoid robotics service provider, it used the live setting to show that embodied intelligence can operate in an actual service environment.
WAVE and full-body motion control split AI tasks from physical movement
Matrix Robotics’ integrated stack combined its independently developed WAVE Embodied AI Foundation Model with a general-purpose full-body motion control system and high-degree-of-freedom dexterous hands. In the system, WAVE handles perception, reasoning and high-level motion planning, while the full-body motion control system coordinates the robot’s physical movements. The company also described long-term testing across multiple scenarios as supporting stable performance for frequent and continuous service at the venue.
MATRIX-3 executed a multi-step autonomous coffee-service workflow
In the dining-area demonstration, MATRIX-3 performed a standardised coffee-service process end to end, including identifying and picking up an empty coffee cup, calculating the force needed to press the coffee machine controls, and delivering a finished hot beverage to guests. The robot planned a route to the guest and delivered the drink steadily as part of a continuous, multi-step workflow. This was presented as an example of multitask processing and generalisation across different movements within a real service task.
Force-control reasoning was used to reduce risks during physical handling
The WAVE model incorporated physical force-control reasoning so the system could adjust grip strength in real time. This capability was described as reducing the risk of cups slipping or hot beverages spilling during live service. The company framed these elements—real-time adjustment during physical interaction and stable multi-step execution—as evidence that embodied AI can be deployed for commercial service applications beyond a one-off lab demonstration.

Reading history

Hide →

This history is cleared when the page is closed.

Can embodied AI move from demos to real service use?

Loading

Was this article helpful?

Helpful

Not helpful

Report an issue

1

0

0

Share with friends