Mobile system for indoor 3-D mapping and creating virtual environments

• US 6,009,359 A
• Filed: 09/18/1996
• Issued: 12/28/1999
• Est. Priority Date: 09/18/1996
• Status: Expired due to Term

• Current Assignee: National Research Council Canada
• Original Assignee: National Research Council Canada
• Inventors: Boulanger, Pierre, El-Hakim, Sabry F.
• Primary Examiner(s): MARQUIS, LISSI MOJICA
• Application Number:
  US08/714,325
• Time in Patent Office: 1,196 Days
• Field of Search: 348/119, 348/135, 348/139, 348/142, 701/25, 701/28
• US Class Current: 701/28
• CPC Class Codes: G01C 11/02 Picture taking arrangements specially adapted for photogrammetry or photographic surveying, e.g. controlling overlapping of pictures

Abstract

A mobile mapping system is designed to generate a geometrically precise three-dimensional detailed model of an unknown indoor environment. There is no limit on the distance the mobile system may travel, nor on the type of details that can be mapped. The same design is used for environments ranging from simple office hallways to long winding underground mine tunnels. Surfaces and features can be mapped from images acquired by a unique configuration of different types of optical imaging sensors and positioning devices. This configuration ensures that all the information required to reconstruct the three-dimensional environment is included in the collected images. Provided a sufficient overlap between two-dimensional intensity images and combining such with three-dimensional range images, the method reconstructs an environment by adjusting a plurality of images simultaneously.

Claims

1. A method of locating at least a surface and modeling the surface comprising the steps of:

a) repeating the following steps;

i) using at least two image capture means having a known spatial relationship therebetween, capturing with each image capture means an image, each of the at least two of image capture means having an image overlapping an image of at least another image capture means from the at least two image capture means;

ii) moving the at least two image capture means to a new location such that each image captured from the new location overlaps at least a previously captured image until a plurality images have been captured of substantially the entire surface;

b) within each image from the plurality of images using a processor to identify distinguishable features;

c) using a processor, aligning identified features in overlapped images using the known spatial relationship between the at least two image capture means as constraints to determine correlations of spatial relations for overlapped images; and

d) using a processor, locating the at least a surface and modeling the at least a surface by mapping the images onto a grid in three-dimensional space and determining the surface that substantially best approximates the correlations and is within the constraints.

2. A method of locating at least a surface and modeling the surface as defined in claim 1 wherein images captured from a same platform location by an image capture means and by adjacent image capture means overlap.

3. A method of locating at least a surface and modeling the surface as defined in claim 2 wherein images captured from a same platform location by an image capture means, by adjacent image capture means, and by image capture means adjacent the adjacent image capture means overlap.

4. A method of locating at least a surface and modeling the surface as defined in claim 2 wherein at least an image capture means is a range image capture means.

5. A method of locating at least a surface and modeling the surface as defined in claim 2 wherein at least an image capture means is an intensity image capture means.

6. A method of locating at least a surface and modeling the surface comprising the steps of:

(a) providing intensity images from a plurality of image capture means having a known spatial relation said images overlapping in part;

(b) providing range images from at least a range image capture means said images overlapping the intensity images at least in part and overlapping at least some other range images in part;

(c) providing registration information in dependence upon known spatial relations between each of the plurality of the image capture means and the at least a range imaging means;

(d) extracting distinguishable features from each intensity image and each range image;

(e) using a processor, correlating the intensity images and the range images in dependence upon the registration information and extracted features located within a plurality of intensity and range images; and

(f) using a processor, locating at least a surface and constructing a model in dependence upon the intensity images and the range images by overlaying at least some intensity images onto the surface in a fashion that best approximates the correlations.

7. A method of locating at least a surface and modeling the surface as defined in claim 6 further comprising the steps of providing movement information in dependence upon a dead-reckoning means and determining initial estimates of location for each image in dependence upon the movement information.

8. A method of locating at least a surface and modeling the surface as defined in claim 7 wherein the model construction is performed in further dependence upon computed vehicle positions.

9. A system for modeling three dimensional space comprising:

(a) a movable platform;

(b) a plurality of image capture devices coupled to said movable platform in known locations and for capturing a plurality of intensity images at least some of the images overlapping each other;

(c) at least a range image capture device coupled to said movable platform in at least a known location and for capturing a plurality of range images;

(d) a memory for receiving the captured intensity images and the captured range images and for storing information indicative of a spatial relation between the captured intensity images and the captured range images and for storing the images; and

(e) a processor coupled to the memory for correlating the plurality of stored intensity images, the plurality of stored range images, the known locations of the intensity image capture devices, and the known location of the range image capture devices and generating a model in dependence upon the correlation.

10. A system for modeling three dimensional space as defined in claim 9 further comprising a dead-reckoning means.

11. A system for modeling three dimensional space as defined in claim 10 wherein said dead reconing means comprise wheel encoders.

12. A system for modeling three dimensional space as defined in claim 9 further comprising an inertial navigation system.

13. A system for modeling three dimensional space as defined in claim 9 wherein each of said plurality of intensity image capture devices are for capturing intensity images that overlap intensity images captured by adjacent intensity image capture devices.

14. A system for modeling three dimensional space as defined in claim 10 wherein the intensity images overlap the at least a range image captured at a same movable platform location.

15. A system for modeling interior three dimensional spaces as defined in claim 10 wherein the plurality of intensity image capture devices is formed of at least 4 intensity image capture devices linearly arranged along one side and at least 2 intensity image capture devices along at least another side.