BMW Group Plant Landshut is using Xsens motion capture to develop software and data infrastructure for AI-supported humanoid robotics in component production. The goal is to identify where humanoid robots can deliver reliable value in real manufacturing environments.
As part of this work, BMW Group is creating a modular robotics ecosystem that combines traditional programming with Vision-Language-Action (VLA) models. To support this, BMW Group collects data from real production environments and integrates it into simulation, motion planning, and robot training workflows before new capabilities are introduced into production.
Alongside these simulations, engineers use demonstration-based learning, where human operators perform manufacturing tasks that are recorded and converted into reusable robot behaviors, or "policies." These learned behaviors can then be adapted across different industrial tasks
Capturing human expertise
Xsens Link captures precise full-body motion during these demonstrations. This includes how operators reach, bend, turn, and coordinate their movements while completing a task.
Motion data can support simulation, motion planning, retargeting, and robot training, helping developers translate human expertise into robot behavior in real industrial environments.
From demonstration to deployment
BMW Group Plant Landshut’s work illustrates an important shift in humanoid robotics. Progress increasingly depends not only on the robot itself, but also on the infrastructure surrounding it: reliable training data, simulation environments, learning models, motion-planning tools, and scalable software platforms.
Human demonstration data provides a connection between production expertise and these emerging robotics systems. By capturing how experienced operators move and interact with their environment, Xsens Link helps turn real human movement into structured data that can support the development of practical industrial robot skills.
Read the full BMW Group announcement.
