Sensor Integration and Filtering

Sensor integration describes the combination of two or more navigation sensors to achieve better results than using a single sensor. At the Institute of Geodesy, this is done using Kalman and particle filters. There are multiple reasons for sensor integration:
  • The estimation of all required parameters (e.g.: GNSS receivers cannot provide (all) attitude parameters).
  • The improvement of the accuracy, reliability and integrity compared to the results of a single sensor.
  • The increase of the update frequency of the solution.
The most common combination is the integration of an IMU (Inertial Measurement Unit) and a GNSS (Global Navigation Satellite System) receiver because of their complementary properties. The optimal filtering of the different sensors includes an additional dynamic model (model of the expected movement of the vehicle) and a time-dependent noise. This optimal filter is implemented in several research projects as Kalman filter or particle filter. At the Institute of Geodesy, the following sensors are combined in various ways:
  • GNSS
  • Inertial Sensors (strapdown approach and pedestrian navigation approaches)
  • Magnetometers
  • Barometer
  • Odometry (e.g. directly from the car via CAN bus)
  • WiFi data
  • BLE data (Bluetooth Low Energy)
Related Projects: OMOSA, CPS, UPIC, ISA

Pedestrian Navigation

Schematic figure showing the 3 different layers maps, position filtering and GNSS satellites for pedestrian navigation
© ifG, TU Graz
Where navigation applications were historically tailored for ships, planes and trucks, today cars as well as pedestrians benefit and use navigation excessively. Unlike in other navigation applications, pedestrians happen to be in adverse conditions for navigation signals and their dynamics are unpredictable. Related Projects: SIPAS, PosCity, LOBSTER

Automotive Navigation

This picture shows a car carrying the platform of the institute with 4 GNSS antennas and receivers and different kinds of inertial measurement units
© ifG, TU Graz
Automotive navigation represents a major backbone of the aspiring field of autonomous vehicles and autonomous driving. Being in knowledge of a vehicle’s position is crucial for many applications; however, the requirements for integrity, availability and accuracy are high due to safety-related aspects. Those requirements are met by means of sensor fusion, whereas internal as well as external sensors have to be taken into account. With an increase of the network capabilities of vehicles, automotive navigation is further augmented by interconnection of vehicles (Car2Car) and the communication with infrastructure such as traffic lights (Car2X). At the Institute of Geodesy, numerous projects were executed to determine accurately and reliably the dynamic state (position, velocity, accelerations, attitude, turn rates) in post-processing and real-time. Therefore, GNSS, INS, magnetometer, odometry and map data were used. Also, projects in the field of cooperative driving are/were executed. To evaluate the quality of the various positioning systems, the institute is in possession of a highly-priced inertial measurement unit called iMAR iNav RQH with ring laser gyroscopes and high-precision accelerometers. This is unique in the cientific field in Austria and Europe. Along with many other sensors and up to 5 GNSS antennas, this IMU may be attached to a tailored and calibrated roof mounting and can be used to evaluate the performance of any navigation sensor. Related Projects: UPIC, ISA, NETI, CrashPos

UAV Navigation

UAV (Unmanned Aerial Vehicles) are a current issue for very different applications. For these systems, important parameters like the position, velocity and attitude are provided by navigation. However, the particular requirements for the navigation sensors are very high. RPAS offer only a limited transport capacity concerning volume as well as weight, hence there are also exact requirements for the size and weight of the sensors. Furthermore, to offer cheap RPAS flights for civil applications, only low-cost sensors are eligible. Nevertheless, the system should work as accurate and reliable as possible, especially when it is coming to BVLOS applications (beyond visual line of sight). The Institute of Geodesy has developed algorithms for accurate and reliable state estimation of UAVs using GNSS (SPP, DGPS, PPP), inertial navigation, magnetometer and barometer. These algorithms are continuously extended and improved. Related Projects: OMOSA

Indoor Positioning

This figure shows a map of a building and a estimated trajectory using indoor positioning sensors
© ifG, TU Graz
The majority of navigation modules today are based on radio signals, which can get distracted by constructions easily and mostly do not work inside buildings. The absence of globally available navigation signals makes positioning indoors a real challenge. Related Projects: NETI, SIPAS, LOBSTER

Routing and Guidance

Manfred Wieser
Steyrergasse 30/III
8010 Graz
Tel: +43/316/873-6348
Fax: +43/316/873-8888