Low cost relative GNSS positioning with IMU integration

In recent years, automotive industry has put much effort into developing intelligent active safety systems to assist the driver. Many such systems depend on relative positioning of vehicles in close proximity. Currently laser, radar and camera sensors are used in mass production while Global Navigation Satellite Systems (GNSS) are often applied as a reference during verification. This thesis evaluates methods for low cost, single frequency GNSS relative positioning which can be applied in verification of active safety systems. Since GNSS sensors operating in the same area are affected by similar errors, the relative accuracy is higher than the absolute for short distances. One of the proposed methods consists of two separate loosely coupled GNSS and inertial measurement unit (IMU) filters for absolute positioning, sharing only the absolute position of each vehicle in order to calculate the distance between them. The second method utilizes a tightly coupled filter where GNSS raw data and IMU measurements of two vehicles are used in a single filter which estimates a relative baseline rather than absolute positions. Both proposed implementations are applying nonlinear Kalman filtering and smoothing. The tightly coupled approach provides superior results in most scenarios. Because the algorithms can be implemented without smoothing and the IMU sensors already exist in modern vehicles, the proposed low cost system can serve as a basis for a real time implementation to support active safety functions.