Method and apparatus for estimating position

Patent metadata

Field	Value
Publication number	US-9386209-B2
Application number	US-201414336415-A
Country	US
Kind code	B2
Filing date	Jul 21, 2014
Priority date	Jan 24, 2014
Publication date	Jul 5, 2016
Grant date	Jul 5, 2016

Abstract

Provided is a position estimation apparatus including: a position estimation unit configured to estimate a position of a moving object driven on an arbitrary lane according to an image frame captured by a stereo camera provided in the moving object; a posture information acquisition unit configured to obtain first posture information of the moving object from the estimated position and second posture information of the moving object at a point of time of the stereo camera capturing the image frame; a position correction unit configured to calculate: a probability distribution for a current position of the moving object by using the first posture information and the second posture information, and configured to correct a first probability of the probability distribution for the current position as a corrected position of the moving object; and a composite probability distribution based on a probability of a lane section of the arbitrary lane from a lane probability distribution chart and the corrected position output from the first position correction unit and configured to re-correct a second probability of the composite probability distribution as a final position of the moving object.

First claim

What is claimed is:

1. A position estimation apparatus comprising one or more processors configured to implement:
a position estimation unit configured to estimate a position of a moving object driven on an arbitrary lane according to an image frame captured by a stereo camera provided in the moving object;
a posture information acquisition unit configured to obtain first posture information of the moving object from the estimated position and second posture information of the moving object at a point of time of the stereo camera capturing the image frame;
a position correction unit configured to calculate a first probability distribution for a position of the moving object by using the first posture information and the second posture information, configured to correct the estimated position based on the first probability distribution, configured to calculate a composite probability distribution using the first probability distribution and a second probability distribution which shows probabilities of the moving object being positioned in lane sections of the arbitrary lane from a lane probability distribution chart in which a lane width of the arbitrary lane is divided into a plurality of sections, and configured to re-correct the corrected estimated position based on the composite probability distribution.

2. The position estimation apparatus of claim 1 , wherein the position estimation unit comprises:
an extraction unit configured to extract a feature point from each of first and second image frames;
a matching unit configured to match the feature point of the second image frame to a corresponding feature point of the first image frame;
a calculation unit configured to calculate a three-dimensional (3D) coordinate of the matched feature point; and
a clustering unit configured to generate at least one motion cluster by grouping a motion vector representing a 3D coordinate space variation between the first image frame and the second image frame, based on a space correlation chart on a feature space;
a noise removing unit configured to remove noise from each motion cluster; and
a position estimation unit configured to estimate the position of the moving object by tracing relative positions of the first and second cameras calculated for the at least one motion cluster.

3. The position estimation apparatus of claim 1 , wherein the second probability distribution shows that a probability of a position of the moving object varies with the lane sections.

4. The position estimation apparatus of claim 1 , wherein the position correction unit is configured to correct the estimated position to have a largest probability in the first probability distribution, and re-correct the corrected estimated position
to have a largest probability in the composite probability distribution.

5. The position estimation apparatus of claim 1 , wherein the posture information acquisition unit is configured to obtain the first posture information including x, y, z, roll, yaw, and pitch of the moving object from the estimated position.

6. The position estimation apparatus of claim 1 , wherein the posture information acquisition unit is configured to obtain the second posture information including, x, y, z, roll, yaw, and pitch of the moving object from position information of the moving object.

7. The position estimation apparatus of claim 1 , wherein the posture information acquisition unit comprises an inertial measurement unit (IMU).

8. The position estimation apparatus of claim 1 , wherein the position correction unit is configured to calculate the first probability distribution by using a factor graph algorithm which uses the first posture information and the second posture information as inputs.

9. The position estimation apparatus of claim 1 , further comprising:
a memory configured to store a road information map and a probability distribution of lane sections of each of lanes included in the road information map.

10. The position estimation apparatus of claim 1 , wherein the second probability distribution is configured to be set so that a central section of the arbitrary lane is set to have largest probability, and probability in the second probability distribution is set to decrease toward an external end of the arbitrary lane.

11. A position estimation method comprising:
estimating a position of a moving object driven on an arbitrary lane from an image frame captured by a stereo camera provided in the moving object;
obtaining first posture information of the moving object from the estimated position;
obtaining second posture information of the moving object at a point of time of the stereo camera capturing the image frame;
calculating a first probability distribution for a position of the moving object by using the first posture information and the second posture information;
correcting the estimated position based on the first probability distribution; and
calculating a composite probability using the first probability distribution and a second probability distribution which shows probabilities of the moving object being positioned in lane sections of the arbitrary lane from a lane probability distribution chart in which a lane width of the arbitrary lane is divided into a plurality of sections; and
re-correcting the corrected estimated position based on the composite probability distribution.

12. The position estimation method of claim 11 , wherein the estimating the position comprises:
extracting a feature point from each of first and second image frames;
matching the feature point of the second image frame to a corresponding feature point of the first image frame;
calculating a three-dimensional (3D) coordinate of the matched feature point;
generating at least one motion cluster by grouping a motion vector representing a 3D coordinate space variation between the first image frame and the second image frame, based on a space correlation chart on a feature space;
removing noise from each motion cluster; and
estimating the position of the moving object by tracing relative positions of the first and second cameras calculated for the at least one motion cluster.

13. The position estimation method of claim 11 , wherein the second probability distribution shows that a probability of a position of the moving object varies with the lane sections.

14. The position estimation method of claim 13 , further comprising:
storing a road information map and a probability distribution of lane sections of each of lanes included in the road information map.

15. The position estimation method of claim 11 , wherein the correcting the estimated position comprises correcting the estimated position to have a largest probability in the first probability distribution, and wherein the re-correcting the corrected estimated position comprises re-correcting the corrected estimated position to have a largest probability in the second probability distribution.

16. The position estimation method of claim 11 , wherein the obtaining the first posture information comprises obtaining the first posture information including x, y, z, roll, yaw, and pitch of the moving object from the estimated position.

17. The position estimation method of claim 11 , wherein the obtaining the second posture information comprises obtaining the second posture information including, x, y, z, roll, yaw, and pitch of the moving object from position information of the moving object.

18. The position estimation method of claim 11 , wherein the correcting the estimated position comprises calculating the fir

Assignees

• Hanwha Techwin Co Ltd

Inventors

• Kim Dong Shin
• Park Jin Suk

Classifications

• G06V10/809

  of classification results, e.g. where the classifiers operate on the same input data · CPC title

• H04N23/57

  Mechanical or electrical details of cameras or camera modules specially adapted for being embedded in other devices · CPC title

• G06T7/73 Primary

  using feature-based methods · CPC title

• G06F18/254

  of classification results, e.g. of results related to same input data · CPC title

• G06T2207/20076

  Probabilistic image processing · CPC title