Experimental system for validating GPS/INS integration algorithms

Today most civil and military navigation systems are based on or include GNSS (Global Navigation Satellite System). In environments where buildings or foliage can block or seriously impede the propagation of the GNSS signal, it is essential to aid the GNSS by using different kinds of sensors with complementary properties for robustness and redundancy. Such a sensor system is the INS (Inertial Navigation System). The drawback of traditional advanced high performing and robust military navigation systems is that they are expensive, bulky and power consuming. By integrating a GPS receiver and a MEMS (Micro Electro Mechanical System) based IMU (Inertial Measurement Unit) one can achieve a navigation system of small size and weight, with modest power consumption and cost. However, the error characteristic of the MEMS sensors is often highly non-linear and temperature dependent. To achieve the desired accuracy it is therefore crucial to determine and model the dominating errors and analyzing their effects in navigation applications.

The work in this master thesis mainly consists of the design and implementation of an experimental platform for logging navigation data. This data is then used for validation and evaluation of robust navigation algorithms using cheap sensors. The report first briefly describes the theory of integration of GPS and INS. Then the implemented test equipment and the used navigation sensors are presented.

Experiments were conducted in both high- and low-dynamic environments, using a roller coaster and a car respectively. Two different integration algorithms, tight and loose integration, are validated by using logged experimental data from the low dynamic car case. Laboratory tests have been performed for the MEMS IMU (MICRO ISU BP3010) to determine its deterministic and stochastic errors. The tests consisted of drift test, gyro-turn-table tests and up-down tests. Both spectral analysis and Allan variance analysis has been used and compared while determining the stochastic errors.

Author: Hjortsmarker, Niklas

Source: Lulea University of Technology

Download Link: Click Here To Download This Report (PDF)

Reference URL: Visit Now

Leave a Comment