Industrial
Metaverse
Merging physical and virtual space
AI Potential and Data Generation in the Industrial Metaverse: From Real-World Testing to Practical Application
The Industrial Metaverse is more than just a simulation; it is the birthplace of Physical AI. For artificial intelligence to not only analyze data but also control physical machines and manipulate robotic arms, it must understand the world.
This is where the work at the ARENA2036 real-world lab comes in. As an innovation center for the Industrial Metaverse, we focus on the critical interface between virtual training and physical execution.
Our focus is on Industrial Perception: visual perception as a key technology. Without precise "eyes," there can be no intelligent actions. We develop pipelines that use synthetic data to teach robust AI models to see, so they can operate safely in real-world factory environments.
Our focus areas in the real-world laboratory
At the ARENA2036 real-world lab, we make the Industrial Metaverse tangible. AI is a key enabler for bridging the gap between the real and virtual worlds. We are exploring the generation and use of synthetic data as a step toward physical intelligence: the goal is to enable machines to “understand” the physical world and interact with it based on this data.
Sim2Real: The basis for physical AI
How does a robot in the metaverse learn to grasp a real screw? We’ll show you how we bridge the gap between reality and simulation and translate virtual knowledge into physical skills.
The Data Factory: Scalable Perception
Data is the fuel. We offer a glimpse into our automated pipeline, which generates tens of thousands of validated training images overnight.
Guidelines: Science meets practice
Based on our experiments, we present a blueprint for datasets that deliver real-world performance.
IRIS Benchmark
New challenging dataset for industrial applications. IRIS is an open dataset for validating perception algorithms using real, standardized industrial parts.
Brownfield & GenAI: The Future of Existing Factories
Physical AI requires digital maps. We are comparing methods such as Gaussian splatting to efficiently adapt existing factories for the Industrial Metaverse.