Apparatus and method for updating detailed map

1. An apparatus for updating a high definition map, the apparatus comprising:

a processor; and

a memory storing instructions thereon, the instructions when executed by the processor cause the processor to:

capture two-dimensional images at a plurality of different locations by using a camera mounted on a vehicle;

determine a moving trajectory of the camera for capturing the two-dimensional images by at least comparing pixel locations of a first landmark other than lane markings within the two-dimensional images;

generate a three-dimensional local landmark map corresponding to the moving trajectory of the camera by using the first landmark;

generate a second local landmark map by estimating, based on surface information of a road in the high definition map, a three-dimensional position of a second landmark of a lane marking around the moving trajectory of the camera; and

update the high definition map based on the local landmark map.

2. A method of updating a high definition map, the method comprising:

capturing two-dimensional images at a plurality of different locations by using a camera mounted on a vehicle;

determining a moving trajectory of the camera for capturing the two-dimensional images by at least comparing pixel locations of a first landmark other than lane markings within the two-dimensional images;

generating a three-dimensional first local landmark map corresponding to the moving trajectory of the camera by using the first landmark;

generating a second local landmark map by estimating, based on surface information of a road in the high definition map, a three-dimensional position of a second landmark of a lane marking around the moving trajectory of the camera; and

updating the high definition map based on the second local landmark map.

3. The method of claim 2 , wherein

wherein the generating the three-dimensional first landmark map includes:

capturing the first landmark at a first location to obtain a first image,

capturing the first landmark at a second location to obtain a second image,

determining a first vector passing, from the first location, through a first pixel X 1 corresponding to the first landmark on the first image,

determining a second vector passing, from the second location, through a second pixel corresponding to the first landmark on the second image,

estimating an intersection of the first vector and the second vector, and

determining a three-dimensional position of the first landmark based on the estimated intersection.

4. The method of claim 2 , wherein the moving trajectory of the camera is determined further by using at least one of an Inertial Navigation System (INS) and a Real Time Kinematic (RTK) GPS.

5. The method of claim 2 , wherein the updating the high definition map includes:

dividing, into grids, each of the local landmark map and an area of the high definition map corresponding to the local landmark map;

comparing each grid of the local landmark map with each grid of the area of the high definition map; and

determining whether to update the high definition map based on a result of the comparison.

6. The method of claim 2 , wherein the updating the high definition map includes:

fitting the second landmark into the second local landmark map based on the estimated three-dimensional position of the second landmark; and

updating the high definition map based on the second local landmark map where the second landmark is fitted.

7. The method of claim 6 , wherein the fitting the second landmark includes:

dividing the second local landmark map into grids; and

fitting the second landmark by a polynomial using adjacent grids among grids corresponding to the estimated position.

8. The method of claim 2 , wherein the generating the second local landmark map includes:

matching the high definition map with the first local landmark map; and

estimating the three-dimensional position of the second landmark in the matched first local landmark map.

9. The method of claim 8 , wherein the matching the high definition map with the first local landmark map includes:

detecting a correspondence between landmarks in the high definition map and the first local landmark map; and

verifying the detected correspondence based on a positional relationship of at least three landmarks in each of the high definition map and the first local landmark map corresponding to each other.

10. The method of claim 8 , wherein the estimating the three-dimensional position of the second landmark includes:

dividing an area corresponding to the moving trajectory of the camera in the high definition map into a plurality of grid planes;

acquiring a vector from an origin of a camera coordinate system to a pixel corresponding to the second landmark in the two-dimensional image; and

estimating the three-dimensional position of the second landmark based on intersection points of straight lines determined from the acquired vector and the plurality of the grid planes.

11. The method of claim 10 , wherein the dividing the area corresponding to the moving trajectory of the camera into a plurality of the grid planes includes:

determining, for a first grid plane including at least four position information among a plurality of the grid planes, an equation of the first grid plane based on the at least four position information; and

determining, for a second grid plane including at most three position information among a plurality of the grid planes, the equation of the first plane adjacent to the second grid plane as an equation of the second grid plane.

12. The method of claim 8 , wherein the estimating the three-dimensional position of the second landmark comprises estimating a plurality of three-dimensional positions of the second landmark from a plurality of two-dimensional images acquired by separate cameras, and

wherein the generating the second local landmark map further includes determining validity of the three-dimensional position of the second landmark by accumulating a plurality of the estimated three-dimensional positions.

13. The method of claim 12 , wherein determining validity of the three-dimensional position of the second landmark includes:

accumulating a plurality of the estimated three-dimensional positions of the second landmark in the first local landmark map divided into a plurality of grids; and

determining, among the plurality of grids, a grid where an accumulation value is equal to or higher than a predetermined threshold value as a valid three-dimensional position of the second landmark.

14. A non-transitory computer-readable storage medium including computer executable instructions, wherein the instructions, when executed by a processor, cause the processor to perform a high definition map updating method comprising:

capturing two-dimensional images at a plurality of different locations by using a camera mounted on a vehicle;

determining a moving trajectory of the camera for capturing the two-dimensional images by at least comparing pixel locations of a first landmark other than lane markings within the two-dimensional images;

generating a three-dimensional first local landmark map corresponding to the moving trajectory of the camera by using the first landmark;

generating a second local landmark map by estimating, based on surface information of a road in the high definition map, a three-dimensional position of a second landmark of a lane marking around the moving trajectory of the camera; and

updating the high definition map based on the second local landmark map.

15. The non-transitory computer-readable storage medium of claim 14 ,

wherein the generating the three-dimensional first landmark map includes:

capturing the first landmark at a first location to obtain a first image,

capturing the first landmark at a second location to obtain a second image,

determining a first vector passing, from the first location, through a first pixel X 1 corresponding to the first landmark on the first image,

determining a second vector passing, from the second location, through a second pixel corresponding to the first landmark on the second image,

estimating an intersection of the first vector and the second vector, and

determining a three-dimensional position of the first landmark based on the estimated intersection.

16. The non-transitory computer-readable storage medium of claim 14 , wherein the generating the second local landmark map includes:

matching the high definition map with the first local landmark map; and

estimating the three-dimensional position of the second landmark in the matched first local landmark map.

17. The non-transitory computer-readable storage medium of claim 16 , wherein the matching the high definition map with the first local landmark map includes:

detecting a correspondence between landmarks in the high definition map and the first local landmark map; and

verifying the detected correspondence based on a positional relationship of at least three landmarks in each of the high definition map and the first local landmark map corresponding to each other.

18. The non-transitory computer-readable storage medium of claim 16 , wherein the estimating the three-dimensional position of the second landmark is estimating a plurality of three-dimensional positions of the second landmark from a plurality of two-dimensional images acquired by separate cameras, and

wherein the generating the second local landmark map further includes determining validity of the three-dimensional position of the second landmark by accumulating a plurality of the estimated three-dimensional positions.

19. The non-transitory computer-readable storage medium of claim 16 , wherein the estimating the three-dimensional position of the second landmark includes:

dividing an area corresponding to the moving trajectory of the camera in the high definition map into a plurality of grid planes;

acquiring a vector from an origin of a camera coordinate system to a pixel corresponding to the second landmark in the two-dimensional image; and

estimating the three-dimensional position of the second landmark based on intersection points of straight lines determined from the acquired vector and the plurality of the grid planes.

20. The non-transitory computer-readable storage medium of claim 19 , wherein the dividing the area corresponding to the moving trajectory of the camera into a plurality of the grid planes includes:

determining, for a first grid plane including at least four position information among a plurality of the grid planes, an equation of the first grid plane based on the at least four position information; and

determining, for a second grid plane including at most three position information among a plurality of the grid planes, the equation of the first plane adjacent to the second grid plane as an equation of the second grid plane.