• Carsten Bundgaard Jacobsen
  • Lars Vestergaard Poulsen
  • Rasmus Rødgaard Riisager
This report deals with the following problem formulation: "How can a dead reckoning navigation system consisting of an IMU, compass and distances between GNSS positions, be designed for positioning in pipelines, and is it possible to achieve an accuracy of ± 0.5 meters or better?" This problem formulation is answered by developing a navigation system, consisting of an IMU, compass and GNSS, and developed software (NS-1002). Two methods of positioning are used. One position is achieved by utilizing the IMU and the other position is achieved by a sensor integration of a compass and calculated distances between GNSS positions. Data from these two methods of positioning are then used in a Kalman filter, resulting in final positions of the covered path, which NS-1002 is capable of plotting afterwards. NS-1002 has been developed in Matlab, and features modules for: communication with instruments, collection of data, processing and presentation of data. NS-1002 has been tested by covering various paths, in order to simulate the movement in a piping system during TV inspection. This was done to determine the achievable accuracy from the sensor integration of the selected sensors and positioning method. The final conclusion of various analyses is that it was not possible to achieve the requirement accuracy stated in the problem formulation. This is caused by the IMU, which accumulates large errors over time in positioning context. This is due to the fact that observations from the IMU are not regularly updated with stable orientation values. These values should be measured continuously during the location determination to correct the IMU observations to local level, in order to achieve a reasonable accuracy. Position determination using distance and direction, has proved considerably more robust over time and were generally the deciding factor of the results achieved from Kalman filtering.
Publication date2009
Number of pages196
Publishing institutionAalborg Universitet
ID: 17683570