| While the surface surveying industry is reaping great benefits in the use of specialized GPS system to enhance, facilitate and speed up the acquisition of survey data, the underground mining surveying industry is not as fortunate as Global Positioning Systems (GPS) signals are not able to penetrate the surface of the earth. This project tries to replicate a three dimensional surveying system used at Inco Ltd. that uses an Inertial Navigation System (INS) in combination with a two dimensional laser scanner to produce a three dimensional survey of the underground environment. The current problem lies in identifying how the error accumulation occurs to then develop error re-distribution techniques to increase the accuracy of the survey. Just knowing the final error accumulation is not acceptable. A target accuracy of 1:10,000 was set as a suitable target for the underground mining sector. Tests were performed by moving a vehicle to and from known points, placing a laser plummet directly over points. Specially designed traverses were used to identify the error accumulation based on direction and speed. The results identify a direct relationship between the direction of travel, clockwise (CW) or counter clockwise (CCW), with the direction of error propagation and magnitudes along each axis. Random errors do occur, complicating the error accumulation, and as the traverse becomes larger and less uniform, a periodic error accumulation is noticed. It was discovered that closing a traverse in the CW direction greatly helps in reducing the error accumulation in both Northing and Easting axis, increasing the survey accuracy. Using the INS surveyor in small rectangular traverses of 500m produced an average survey accuracy of 1:14,243 where maximum accuracies of 1:17,112 were produced during CW experiments and as low as 1:10,253 during CCW traversing. During 4km traverses,{09}survey accuracies of over 1:20,000 were produced during eight of nine different experiments. One experimented failed, caused by improper alignment and could be quickly identified by the gross error accumulations. Again the CW traverses produced greater accuracies, with an average of 1:30,000 and a maximum of 1:40,813 when traveling 30km/h. CCW directions produced an average accuracy of 1:25,916. A final test revealed a direct relationship involving the time between Zero Velocity Update Times (ZUPT) and the achievable accuracy, where shorter time between ZUPTs produced greater survey accuracies. This is attributed to the reduction of non-linear error drift accumulation between ZUPTs. While the INS uses complicated methods of error drift reduction, it cannot fully compensate for large error accumulation between ZUPTs. This research thus highly recommends that ZUPTs be performed every 30 to 40 seconds.; An error correction method applying point averaging was developed to determine the position of control points. The first step applies a polynomial regression to each axis, geared to identify and eliminate systematic errors and exposing the random errors. The second step involves averaging each point collected at each control position to increase the precision of each control point. Finally, random errors can be further corrected by applying an error redistribution algorithm that is linearly time distributed between control points. This method of error correction could increase survey accuracies from 1:30,000 to 1:100,000.; Finally, the introduction of a laser scanner allows a true 3-dimensional survey. The laser scanner introduced an additional +/-2cm error. The real problem lies in identifying the time stamping error and its measurement effects. This project used a 10ms resolution for time stamping due to software limitations, which introduces a +/-5cm error while traveling at 30km/h. This error can be greatly reduced by enhancing the program to use much smaller time step resolutions.; This project demonstrates that a Horta INS can provide positional data with enough accu... |
