Insights from the 2026 Humanoid Robot Forum

Research evidence excerpt

Insights from the 2026 Humanoid Robot Forum

humanoid users will share deployment data or keep it private, potentially limiting the

pace of training improvements.

Still, new data will eventually come through, and capabilities will improve as it does. With core

functions such as standing, walking and lifting now being mastered, the focus of training is

shifting to more advanced models. This includes training physical AI at the system level,

enabling groups of robots to operate in coordination, interacting with their environment and

with other robots. At the same time, new questions are emerging on how to train specific skills

effectively while avoiding overfitting.

#2 Integration

As deployments are becoming more common, mainly in manufacturing, industry participants

are starting to recognize that humanoid robot deployment requires an integration strategy.

This involves connecting these systems with existing automation platforms while ensuring

smooth interaction with human workers. In manufacturing in particular, the objective is

increasingly to enable full communication and coordination between humanoid robots and

existing technologies, allowing factories to operate as integrated systems.

To move toward this, potential adopters are starting to recognize the need to define where in

their production processes it makes sense to deploy humanoid robots. The answer depends on

a range of factors, from where they can improve productivity, safety and scalability, to where

they can relieve workers from dull, dirty and dangerous tasks. The emerging view is that

humanoid robots are better suited to tasks requiring high dexterity and flexibility in

environments with significant variability, but where these tasks are located differs across