Landmark, apparatus, and method for effectively determining position of autonomous vehicles

• US 20040062419A1
• Filed: 05/05/2003
• Published: 04/01/2004
• Est. Priority Date: 10/01/2002
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: Son, Young, Kwak, Joo-Young, Roh, Kyoung-Sig, Park, Ki-Cheol
• Granted Patent: US 7,333,631 B2
• Time in Patent Office: Days
• US Class Current: 382/104
• CPC Class Codes: G06V 10/42 Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
• CPC Class Codes: G06V 10/46 Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features colour feature extraction G06V10/56
• CPC Class Codes: G06V 20/10 Terrestrial scenes scenes under surveillance with static cameras G06V20/52; scenes perceived from the exterior of a vehicle G06V20/56; scenes perceived from the interior of a vehicle G06V20/59
• CPC Class Codes: G06V 2201/11 Technique with transformation invariance effect

Abstract

A landmark used to effectively determine the location of an autonomous vehicle, and a self-localization apparatus and method using the landmark are provided. In the self-localization method, first, first and second outer line information and shape information are extracted from a landmark image received from a camera. Next, a projective invariant is calculated from the shape information and stored in a hash table. Thereafter, the calculated projective invariant is compared with reference projective invariants for a plurality of landmarks stored in a predetermined data storage area in the form of a hash table, thereby determining which landmark corresponds to the landmark image. Then, information on the distance and orientation of the determined landmark with respect to the autonomous vehicle is analyzed in response to the first and second outer line information.

Claims

1. A landmark comprising:

a first outer line area used to ascertain the X and Y axes of an image plane upon self-localization of an autonomous vehicle;

a second outer line area used to ascertain a deviation degree between a camera and a landmark upon acquisition of the image plane; and

a shape area representing the unique shape of the landmark.

2. The landmark of claim 1, wherein the first outer line area has a shape of a rectangle.

3. The landmark of claim 1, wherein the second outer line area has a shape of a perfect circle with a predetermined radius.

4. The landmark of claim 1, further comprising a color area formed between the first and second outer line areas and filled with a predetermined color.

5. An apparatus for determining the location of an autonomous vehicle, the apparatus comprising:

a feature data extractor extracting first and second outer line information and shape information from a landmark image received from a camera;

a projective invariant calculator calculating a projective invariant from the shape information;

a data storage unit storing the calculated projective invariant and reference projective invariants for a plurality of landmarks in the form of a hash table;

a hash table search unit determining which landmark corresponds to the landmark image by comparing the calculated projective invariant with the reference projective invariants for the plurality of reference landmarks; and

a distance/orientation information analyzer analyzing information on the distance and orientation of the determined landmark with respect to the autonomous vehicle in response to the first and second outer line information.

6. The apparatus of claim 5, wherein the landmark comprises:

a first outer line area having the first outer line information used to distinguish the X and Y axes of an image plane upon self-localization of the autonomous vehicle;

a second outer line area having the second outer line information used to ascertain a deviation degree between the camera and the determined landmark upon acquisition of the image plane; and

a shape area having the shape information used to distinguish the landmarks from one another.

7. The apparatus of claim 6, wherein the first outer line area has a shape of a rectangle where the X and Y axes cross each other at a right angle.

8. The apparatus of claim 6, wherein the second outer line area has a shape of a perfect circle with a predetermined radius.

9. The apparatus of claim 5, wherein the projective invariant maintains a constant value without being affected by changes in the shape of the landmark image received from the camera.

10. The apparatus of claim 9, wherein when P denotes an object point, and q denotes an image point corresponding to the object point P, the projective invariant is calculated using the following equation:

11. The apparatus of claim 5, wherein the projective invariant calculator divides the outer line of the landmark shape into n sections, obtains coordinates of points that constitute the n sections, and calculates the projective invariants of the coordinates by moving the coordinates by 1/N times the length of the outer line until the coordinates reach their original locations.

12. The apparatus of claim 11, wherein when 1≦

k≦

N, and X(k) and Y(k) denote X- and Y-axis coordinate functions, respectively, of the outer line of a landmark shape, the projective invariant of each of the n sections is calculated using the following equation;

13. The apparatus of claim 5, wherein if the shape of the second outer line is an oval, the distance/orientation information analyzer transforms an equation for a cubic section of the oval into a cubic section equation for a perfect circle and obtains information on the orientation and distance of the determined landmark with respect to the autonomous vehicle from the relationship equation between the two conic section equations.

14. The apparatus of claim 13, wherein, when a vector n normal to an image plane corresponding to the transformed perfect circle, is (0 0 1)^T, a vector ct′

to the center of the perfect circle is (0−

dc/f d)^T, and the difference (c) between the coordinates of the center of the perfect circle and the center of the cubic section extracted from the landmark image is {square root}{square root over ((α

−

1)(γ

+f²))} (where α

is

15. An autonomous vehicle having a self-localization function, comprising:

an image acquiring unit capturing a landmark image received from a camera in a predetermined image format;

a main controller performing a self-localization function and an overall control for operating the autonomous vehicle, in response to the landmark image captured by the image acquiring unit; and

a movement controller controlling the movement of the autonomous vehicle in response to the control of the main controller,

wherein the main controller comprises;

a feature data extractor extracting first and second outer line information and shape information from the landmark image;

a projective invariant calculator calculating a projective invariant from the shape information;

a data storage unit storing the calculated projective invariant and reference projective invariants for a plurality of landmarks in the form of a hash table;

a hash table search unit determining which landmark corresponds to the landmark image, by comparing the calculated projective invariant with the projective invariants for the plurality of reference landmarks; and

a distance/orientation information analyzer analyzing information on the distance and orientation of the determined landmark with respect to the autonomous vehicle in response to the first and second outer line information.

16. A method of determining the location of an autonomous vehicle, the method comprising:

extracting first and second outer line information and shape information from a landmark image received from a camera;

calculating a projective invariant from the shape information and storing the projective invariant in the form of a hash table;

determining which landmark corresponds to the landmark image, by comparing the calculated projective invariant with projective invariants for a plurality of reference landmarks stored in a predetermined data storage area in the form of a hash table; and

analyzing information on the distance and orientation between the determined landmark and the autonomous vehicle in response to the first and second outer line information.

17. The method of claim 16, wherein the landmark comprises:

a first outer line area having the first outer line information used to distinguish the X and Y axes of an image plane upon self-localization of the autonomous vehicle;

a second outer line area having the second outer line information used to ascertain a deviation degree between the camera and the determined landmark upon acquisition of the image plane; and

a shape area having the shape information used to distinguish the landmarks from one another.

18. The method of claim 17, wherein the first outer line area has a shape of a rectangle where the X and Y axes cross each other at a right angle.

19. The method of claim 17, wherein the second outer line area has a shape of a perfect circle with a predetermined radius.

20. The method of claim 16, wherein the projective invariant maintains a constant value without being affected by changes in the shape of the landmark image received from the camera.

21. The method of claim 20, wherein when P denotes an object point, and q denotes an image point corresponding to the object point P, the projective invariant is calculated using the following equation:

22. The method of claim 16, wherein calculating and storing a projective invariant comprises:

dividing the outer line of the landmark shape into n sections;

obtaining coordinates of points that constitute the n sections; and

calculating the projective invariants of the coordinates by moving the coordinates by 1/N times the length of the outer line until the coordinates reach their original locations.

23. The method of claim 22, wherein when 1≦

k≦

N, and X(k) and Y(k) denote X- and Y-axis coordinate functions, respectively, of the outer line of a landmark shape, the projective invariant of each of the n sections is calculated using the following equation;

24. The method of claim 16, wherein analyzing information on the distance and orientation between the determined landmark and the autonomous vehicle comprises:

transforming an equation for a cubic section of an oval into a cubic section equation for a perfect circle if the shape of the second outer line is the oval; and

obtaining information on the orientation and distance of the determined landmark with respect to the autonomous vehicle from the relationship equation between the two conic section equations.

25. The method of claim 24, wherein, when a vector n′

normal to an image plane corresponding to the transformed perfect circle is (0 0

     1)^T, a vector ct′

to the center of the perfect circle is (0 ▪

dc/f d)^T, and the difference (c) between the coordinates of the center of the perfect circle and the center of the cubic section extracted from the landmark image is {square root}{square root over ((α

−

1)(γ

+f²))} (where α

is

26. A computer recording medium which stores a computer program for executing the method of claim 16.