An accurate and robust vehicle localization is the basis for future driver assistance systems, data exchange between vehicles and their environment (Car2X), and autonomous driving. Difficulties arise from poor reception of global navigation satellite signals, which often occurs in urban environments. In order to provide an accurate navigation solution in difficult situations, existing navigation systems are improved and novel algorithms are developed at the ITE.
Source: Forschungsinitiative Ko-FAS
Global satellite navigation systems (GNSS) are widely used for localization of automobiles. The ITE research focus is on signal processing as well as the correction of GNSS raw data.
Besides GPS, the Russian GLONASS and the European Galileo satellite navigation systems are used to increase the number of visible satellites and, therefore, the accuracy of the navigation solution. Further improvements are obtained by using SBAS (Satellite Based Augmentation System) or Differential GNSS.
Moreover, algorithms for improving the robustness of GNSS-aided navigation solutions are developed at the ITE. This comprises the reduction of multipath effects, which yields to significant improvements especially in urban environments.
Sensor Data Fusion
An increase in availability and accuracy of a navigation solution is achieved by fusing measurements of complementary sensors. Typically, an inertial navigation system is extended by fusing the inertial measurement unit (IMU) data with GNSS measurements. GNSS/INS data fusion is reached by a loosely coupled or a tightly coupled system. Moreover, a method to increase the robustness of the navigation system is to use a deeply coupled system with a direct feedback of the navigation solution to the signal tracking of the GNSS receiver.
Besides IMU and GNSS measurements, steering angle and odometer readings, as well as information from cameras and laser rangefinders are used in the data fusion process. In order to optimally use these measurements, methods for processing time-delayed relative state measurements are developed at the ITE.
In addition to data fusion in a standard Kalman filter, advanced decentralized and dynamic filter structures are developed. For example, a federated filter or a segmented navigation filter approach allow for a better fault detection, isolation, and recovery.
Datenfusion of a multisensor system
A GNSS/INS navigation system for automobiles is further improved by including vehicle dynamics constraints. This increases the quality of the navigation solution, especially during GNSS outages. Hence, robustness and availability of the navigation system is improved without requiring additional sensors to be installed in the vehicle.
At the ITE developed Tightly-Coupled GNSS-INS
Ego-Positioning System of the Research Initiative Ko-FAS