Mapping buried utilities in difficult environments

There is a large number of underground utilities buried in urban areas, which is one of the most complex networks in the world. It has been estimated that only 50% of buried utilities are accurately recorded. However, failure to identify accurately the location of existing buried utilities results in numerous practical problems, costs and dangers for utility owners, contractors and road users. The underground utilities positioning accuracy requirement is 100 mm for both the accuracy of positioning system and the accuracy of detection devices. While the accuracy up to 300 mm would be acceptable for many respondents. This aim of this thesis is to research various means of improving the accuracy of positioning systems and the accuracy of detection devices for underground utilities in urban areas. GNSS is mainly used to find and record the position of utilities. However, the performance of GNSS is constrained by an insufficient number of visible satellites, poor satellite geometry and multipath in urban areas. The combination of GNSS systems increases the possible visible satellite number. Moreover, the geometry of satellites will be improved by integrating different GNSS constellations. This thesis evaluates the performance of different GNSS constellations such as GPS, GLONASS, BDS and QZSS and multi-GNSS integration in a controlled environment at UNNC and Ningbo city centre. The results provide evidence that using more than one GNSS constellation will significantly increase the availability of GNSS positions and improve the satellite geometry. There are 75% markers (21 out of 28) on campus of UNNC obtained the positioning error within 10cm either by GPS, BDS or GPS and BDS integration. In Ningbo city centre static test, 47% positions (7 out of 15) obtain ambiguity fixed solutions by GPS and BDS. For the underground utilities detection system, this thesis develops a low-cost IMU and odometer integration system to estimate the position of an approximately 30m long test pipeline. Moreover, a tightly coupled integration between IMU and odometer is developed to decrease error caused by the odometer installation attitude error and scale factor error. Besides this, a novel approach to this application of using a Robust Kalman filter is developed to remove the effect of odometer measurement outliers due to wheel-slip. Compared with the loosely coupled integration method, the use of loosely coupled integration with scale factor correction, tightly coupled integration and tightly coupled with Robust Kalman filter provide a horizontal position improvement of 11%, 41% and 43%, respectively. Similarly, the height accuracy is improved by 14%, 50% and 57% before the wheel-slip. The Wheel-slip leads to wrong odometer measurement that makes the positioning results far away from the truth. After applying Robust Kalman filter, the positioning error is reduced to 0.61 m in the horizontal plane, and 0.11 m in the height. Moreover, if using the forward and backward Kalman filter with known start and end positions, the test pipeline positioning maximum error in height is 4 cm, and the maximum horizontal error is 10 cm.