Case-Study: Semantic Mapping

From Environmental Understanding to Digital Twins

At the Center Connected Industry in Aachen, we have been working together with T-Systems on developing semantic mapping for mobile industrial robots and transport systems. The concept is simple but gets momentum with the usage of 5G and a factory edge infrastructure: combining spatial maps utilized for autonomous navigation with AI-based software modules for object detection and localization.

The mobile robots are equipped with an RGB + depth camera following the cloud robotics concepts and streaming the real-time sensor data to tailored AI services on the edge. The AI service has been trained on a huge industrial dataset that we acquired over the last years. As autonomous robots are able to localize themselves on the map, we could merge the information on detected and localized objects. The result is no longer maps that just differ between free spaces and obstacles. Moreover, we can provide real-time maps that provide a deep understanding of the semantics structure of the environment.