Improved perception capabilities
Thanks to intelligent perception capabilities, the disinfection robots will be able to clean in a targeted manner. A new multimodal 3D sensor from Fraunhofer IOF is used for this purpose. With the help of this sensor, during commissioning, the robots independently recognize all the objects they are to disinfect and the material they are made of. The object recognition system of Fraunhofer IPA and the material recognition system of Fraunhofer IPM evaluate the sensor data using machine learning methods. In this way, they achieve robust recognition, even if the objects look slightly different in each application environment.
A multilayer environment model developed by Fraunhofer IOSB brings together all the necessary information and thus enables the robots to plan cleaning sequences independently. It contains a map of the environment, the position of all objects to be cleaned, and their material. The environment data does not always have to be taught manually. Based on the work of Fraunhofer Italia, it will be possible to automatically load this information from the “Building Information Modeling” (BIM), a digital representation of key building components that already exists for many buildings.
In future, a perception function will also be used in regular operation before individual objects are disinfected: Based on the degree of contamination, the robots should be able to optimize the cleaning process and verify its success. In the project, Fraunhofer FEP is conducting the first fundamental investigations into how this contamination could be detected.