A METHOD AND SYSTEM FOR GEOREFERENCING UNDERGROUND DATA

Abstract

Known georeferencing techniques require input in the form of manually-chosen anchor points or dense surveyed data. The present invention is an improved method and system for georeferencing underground geometric data. The method comprises (a) visiting at least two control points; (b) obtaining information about each of the at least two control points using scanning means; (c) recording the information about the at least two control points into a computer processor; and (d) performing a best-fit transformation to the recorded information. Preferably, the scanning means comprises laser scanners and at least two radio-frequency identification (RFID) tags. However, other technologies, such as retro-reflective LIDAR targets, Wi-Fi access points or bar codes and a bar code reader may also be used. In addition, sonar, radar, flash LIDAR, MEMS LIDAR, or any other similar technology could be used.

Claims

1. A method for georeferencing underground geometric data, the method comprising: (a) visiting at least two control points; (b) obtaining information about each of the at least two control points using data acquiring means; (c) recording the information about the at least two control points into a computer processor; and (d) aligning other scanned data to the recorded data.

2. The method according to claim 1, wherein aligning the other scanned data to the recorded data comprises performing the best-fit transformation on the other scanned data.

3. The method according to claim 1, wherein the best-fit transformation comprises applying a least-squares method to the other scanned data.

4. The method according to claim 1, wherein the data acquiring means comprises at least one laser scanner and at least two radio-frequency identification (RFID) tags.

5. The method according to claim 1, wherein the data acquiring means comprises at least one laser scanner and at least two of the following elements; RFID tags; retro-reflective LIDAR targets; Wi-Fi access points; bar codes and a bar code reader; sonar; calibrated cameras; radar; flash LIDAR; and MEMS LIDAR.

6. The method according to claim 1, wherein the information about the at least two control points comprises the position (X,Y,Z coordinates) of the at least two control points.

7. The method according to claim 1, wherein the computer processor comprises a computer-readable data storage medium.

8. The method according to claim 1, wherein the at least two control points comprise a plurality of control points.

9. Use of the method of claim 1, for rapid surveying of long sections of an underground tunnel.

10. Use of the method of claim 1, for regular scanning of the same area of an underground tunnel over time.

11. Use of the method of claim 1, for feature identification and extraction from at least one of a geological data set, a geotechnical data set, and an infrastructure data set.

12. A system for georeferencing underground data, the system comprising: (a) visiting at least two control points; (b) data acquiring means for obtaining information about each of at least two control points; and (c) computer processor means constructed and arranged to: (i) record the information about the at least two control points; and (ii) align other scanned data to the recorded data.

13. The system according to claim 12, wherein aligning the other scanned data to the recorded data comprises performing the best-tit transformation on the other scanned data.

14. The system according to claim 13, wherein performing the best-fit transformation comprises applying a least-squares method to the other scanned data.

15. The system according to claim 12, wherein the scanning means comprises at least one laser scanner and at least two radio-frequency identification (RFID) tags.

16. The system according to claim 12, wherein the data acquiring means comprises at least one laser scanner and at least two of the following elements: RFID tags; retro-reflective targets; Wi-Fi access points; bar codes and a bar code reader; sonar; calibrated cameras; radar; flash LIDAR; and MEMS LIDAR.

17. The system according to claim 12, wherein the information about the at least two control points comprises the position of the at least two control points.

18. The system according to claim 12, wherein the computer processor means comprises a computer-readable data storage medium.

19. The system according to claim 12, wherein the at least two control points comprise a plurality of control points.

20. Use of the system of claim 12, for rapid surveying of long sections of an underground tunnel.

21. Use of the system of claim 12, for regular scanning of the same area of an underground tunnel over time.

22. Use of the system of claim 12, for feature identification and extraction from at least one of a geological data set, a geotechnical data set, and an infrastructure data set.

Description

DESCRIPTION OF THE DRAWINGS

[0043] Reference is made to the following drawings wherein,

[0044] FIG. 1 is a schematic showing the effect of drift error;

[0045] FIG. 2 shows correspondences between saved snapshots and collected scan data; and

[0046] FIG. 3 illustrates a set of position and headings visited by a laser scanner during a scan.

DETAILED DESCRIPTION OF THE INVENTION

[0047] Referring to FIG. 1, a solid outline representing a top-down view of a portion of a mine is shown. The dashed outline represents a map created using a mobile scanner exhibiting drift error. This map is aligned (georeferenced), but still contains drift error.

[0048] The method of the invention includes two phases: (i) a control point profiling, and (ii) nonrigid transformation. The advantage of nonrigid transformation is that it may be used for georeferencing and/or error correction while rigid transformation may be used for georeferencing only. Each of the above two phases is now described more fully.

[0049] Control point (tag) profiling is the process of visiting each control point with a laser scanner and recording a snapshot of the local area. This part of the process therefore obtains a set of easily-recognizable and globally-positioned landmarks that can be used to georeference subsequently scanned data. For each control point: [0050] 1. a laser scanner is precisely aligned with a control point (e.g., a scanner-mounted laser pointer can be used to position the scanner so that it is directly beneath the control point); [0051] 2. the sensor data observed by the laser scanner is recorded while it is still precisely positioned (often referred to as a snapshot of the local environment) [0052] 3. the sensor data snapshot is saved (e.g., in the form of a tag database that allows easy retrieval of snapshots in any order at a later time).

[0053] Near each control point, an RFID tag is preferably installed which is easily and uniquely recognized when the laser scanner is nearby. The RFID tag's unique ID is included in the snapshot and is used to index the database. The snapshots additionally contain laser data as observed from the precisely-aligned position beneath the control point.

[0054] Later, when all the scan data is acquired for an area, the database can be used to establish correspondences between points in the scan data and points in the real world. Laser data acquired during the scan is compared with saved laser snapshot data to establish precise offsets between laser scanner positions and the known control points. The RFID tags are needed because many areas in the scan may look similar, and installed tags can easily disambiguate them by having a unique serial number.

[0055] While the present invention preferably uses at least one laser scanner and at least two RFID tags, these could be replaced with other technologies to achieve the same result. Other technologies that could be used include Wi-Fi access points (which are also uniquely identifiable), bar codes used with a bar code reader, and numbers painted on a mine wall with a calibrated camera that could recognize them. In addition, retro-reflective LIDAR targets, sonar, radar, flash LIDAR, and MEMS LIDAR, or any other similar technology could be used.

[0056] After the control point (tag) profiling is completed, mobile LIDAR scanning is performed in the area to create 3D point cloud data, then nonrigid registration is performed in which the collected 3D point cloud data is aligned with the control points. In general, the nonrigid registration comprises two steps: a control point association search and registration.

[0057] First, a search is performed to generate the correspondences shown in FIG. 2. As shown in FIG. 2, the local snapshots each store sensor data as acquired while precisely aligned with the control points. After acquiring a mobile scan of the area, correspondences are established between points in that scan and the local snapshots by identifying unique features. These correspondences are then used to relate the known control point coordinates to points on the driven path.

[0058] At each point in time during the scan, if an RFID tag was observed, that tag ID is located within the database. If it is found in the database, this means that the scanner was within read range of the tag at that time, within approximately 5 m. To obtain an even more accurate position relative to the known control point, the laser data acquired at this point in time is compared to that stored in the relevant snapshot using a feature-matching routine based on iterative closest point (ICP; an algorithm used for matching laser scans) to more precisely estimate the relative offset from the known point based on matching the laser data. While ICP is preferred, other feature matching algorithms could be used instead. If successful, a correspondence between the laser scanner's location at this point in time and a known point in real world coordinates is established.

[0059] After searching for associations using the process outlined above, a large set of scanner position and headings, representing the scanner's position and orientation at each point in time (in local coordinates), and a smaller set of correspondences exists. These correspondences establish real-world coordinates for a subset of the scanner position and headings. The next step is registration, which refers to the process of estimating the global position and orientation of the scanner at each point in time, in global (real-world) coordinates, given the information above.

[0060] With the set of correspondences, a best fit transformation is applied to all position and heading data in the set. This, however, while having the effect of bringing position and coordinates into the georeferenced (real-world) frame, does not have any effect on correcting the accumulated drift error. For example, if scanning the straight path from control point A to B yields a path of length 11 m, but the control points are known to be separated by 10 m, a correction should be applied to all position and heading data from the scan to adjust its length to match.

[0061] To both georeference the scan and use the relative information between the control points to correct the error, a nonrigid registration method is used. This begins by using a rigid alignment as described above to approximately align the set of positions and headings (poses). A best-fit transformation is computed by applying a least-squares method to the set of control point correspondences, and this transformation is then applied to each position and heading in the scan path. The result is a new path that is a rotated and translated copy of the original, but not stretched or distorted. As noted above, this represents the original scan data (including drift error) moved into the georeferenced frame, and is a good starting point for the nonrigid (error-correcting) registration process.

[0062] The nonrigid registration step is formulated as an optimization problem that aims to minimize the sum of the squared error of the entire set of poses (i.e. it is a least-squares method). Referring to FIG. 3, a sequence of poses visited by the scanner, with two fixed endpoints is shown. The fixed points 1 and 5 are known by association with surveyed control points, and the remaining position and headings 2, 3, and 4 are known only relative to the fixed endpoints. The deltas .sub.i,j are the relative transitions from position and heading i to position and heading j as measured by the scanner.

[0063] The goal is to fix poses 1 and 5 in place, while allowing the others to move freely into their best fit positions. This is done by defining an objective function:

[00001]$\begin{array}{cc}f=\underset{n}{.Math.}.Math.{\left({}_{i,j}-\right)}^2,& \left(1\right)\end{array}$

which is the sum of the squared difference between each measured delta .sub.i,j and , which is the delta that would be measured with poses i and j in their current positions. By varying the set of positions and headings (except the fixed points) and attempting to find the set which minimizes Equation (1), a set of positions and headings that is both fixed in place by the known coordinates, and simultaneously is the best fit to the measured data is obtained.

[0064] With the optimization problem defined, a heuristic method is used to find a set of position and headings that satisfies the constraints. This position and heading set represents the actual path travelled by the laser scanner in the external coordinate system. Knowing the path travelled by the laser scanner, the laser measurements are then projected into this space, resulting in a georeferenced point cloud.

[0065] The method and system of the present invention may be used for rapid surveying of long sections of underground tunnels (i.e., kilometers), allowing for analyses such as dimensioning (tunnel measurements), horizontal and/or vertical clearance calculations, clash detection for new equipment and/or infrastructure; regular scanning of the same area of an underground tunnel over time, and comparisons of the data sets to detect physical changes in tunnel condition, shape, and profile (e.g. physical changes due to tunnel wall damage from equipment over time, tunnel convergence over time, rock support degradation, etc.); or feature identification and extraction from at least one of a geological data set, a geotechnical data set, and an infrastructure data set.

[0066] The scope of the claims should not be limited by the preferred embodiments set forth in the examples, but should be given the broadest interpretation consistent with the description.