It is well known that automation will play an increasingly important role in the future due to demographic developments and other factors. We have developed a sensor system with which existing vehicles can be retrofitted without great effort and cost to enable a higher level of automation. This system consists of an interconnected network of radar and inertial sensors, as well as data processing algorithms exploiting their data. The radar sensors offer privacy preserving all weather data for range and velocity measurements. On the other hand, inertial measurement units offer additional information for orientation as well, but are prone to drifts. Using these two streams of data in a signal processing and learning framework offers significant improvements in the quality of localisation and this has been pursued in this project. The streamlined localization and mapping system makes it possible to ensure autonomous behavior of existing solutions by minimizing dependencies on other systems. The network technology makes the system adaptable to any shape and size of vehicle. The independence from wheel encoders offers the possibility of a plug-and-play solution without major installation and configuration efforts on the existing system.
