Method for extracting curb of road using laser range finder and method for localizing of mobile robot using curb informaiton of road

Patent metadata

Field	Value
Publication number	US-2016011594-A1
Application number	US-201414521699-A
Country	US
Kind code	A1
Filing date	Oct 23, 2014
Priority date	Jul 9, 2014
Publication date	Jan 14, 2016
Grant date	—

Abstract

The invention relates to a method for extracting a curb of a road using a laser range finder and a method for localizing of a mobile robot using curb information of a road. The method for extracting the curb of the road using the laser range finder includes extracting a road surface and line segments from scan data of the laser range finder, extracting a plurality of curb candidate line segments among the line segments on the basis of an angle between the road surface and the line segment, extracting a plurality of curb candidates having a plurality of curb properties, wherein each of the plurality of curb candidates is generated by combining the couple of the curb candidate line segments, and applying the plurality of the curb candidates to a Kernel Fisher Discriminant Analysis to extract a final curb.

First claim

1 . A method for extracting a curb of a road using a laser range finder comprising following steps:
(a) extracting a road surface and line segments from scan data of the laser range finder; (b) extracting a plurality of curb candidate line segments among the line segments on the basis of an angle between the road surface and the line segment; (c) extracting a plurality of curb candidates having a plurality of curb properties, wherein each of the plurality of curb candidates is generated by combining the couple of the curb candidate line segments; and (d) applying the plurality of the curb candidates to a Kernel Fisher Discriminant Analysis to extract a final curb.

2 . The method for extracting the curb of the road using the laser range finder according to claim 1 , wherein the step (c) comprises following steps:
(c1) extracting a right curb candidate line segment and a left curb candidate line segment from the plurality of curb candidate line segments; (c2) calculating a difference between a distance to the extracted road surface and a distance to the right curb candidate line segment and/or the left curb candidate line segment; (c3) calculating an angular difference between the right curb candidate line segment and the left curb candidate line segment; (c4) calculating a difference between a road width and a distance between the right curb candidate line segment and the left curb candidate line segment; and (c5) extracting the curb candidates having the curb properties with values calculated in the step (c2), (c3), and (c4).

3 . The method for extracting the curb of the road using the laser range finder according to claim 2 , wherein vector data are extracted as the curb candidates and wherein elements of the vector date comprise the values calculated in the step (c2), (c3), and (c4).

4 . The method for extracting the curb of the road using the laser range finder according to claim 3 , wherein the step (d) comprises following steps:
(d1) setting-up a discriminant function via applying a plurality of training data which are classified as a curb class and a non-curb class to the Kernel Fisher Discriminant Analysis; (d2) applying the curb candidate to the discriminant function to classify the curb candidate as one of the curb class and the non-curb class; and (d3) extracting the curb candidate, which is classified as the curb class, as the final curb in the step (d2).

5 . The method for extracting the curb of the road using the laser range finder according to claim 4 , wherein the discriminant function in the step (d1) is generated as an eigenvector maximizing an object function in a formula which is defined as

J


(
α
)

=

α
T



S
B
Φ


α

α
T



S
W
Φ


α

(here, J(α) is the object function, S B Φ is a between-class variance matrix, S W Φ is a within-class variance matrix, and α is the eigenvector).

6 . The method for extracting the curb of the road using the laser range finder according to claim 5 , wherein in the step (d2), Mahalanobis distance between each class and the curb candidate is calculated, and the curb candidate is classified as one of the curb class and the non-curb class which possesses the smaller Mahalanobis distance.

7 . The method for extracting the curb of the road using the laser range finder according to claim 6 , wherein the Mahalanobis distance is calculated by

d


(

y
test

,

μ
c
Φ

)

=

(

y
test

-

μ
c
Φ

)

2

(

σ
c
Φ

)

2

(here, μ c Φ and σ c Φ are the mean and the standard deviations, respectively, of the training data, the curb class, and the non-curb class in a one-dimensional solution space).

8 . A method for localizing of a mobile robot using curb information of a road comprising following steps:
(A) predicting a position of the mobile robot by applying odometry information of the mobile robot to an Extended Kalman Filter; (B) applying a measurement value measured by a DGPS module installed on the mobile robot and an inertial sensor and the position of the mobile robot predicted in the step (A) to the Extended Kalman Filter to correct the position of the mobile robot predicted in the step (A); (C) extracting curb information of the road on the basis of scan data of a laser range finder which is installed on the mobile robot; and (D) applying the extracted curb information and the position of the mobile robot which is corrected in the step (B) to the Extended Kalman Filter to correct the position of the mobile robot which is corrected in the step (B).

9 . The method for localizing of the mobile robot using curb information of the road according to claim 8 , wherein in the step (C), the method of extracting curb information comprises:
(a) extracting a road surface and line segments from scan data of the laser range finder; (b) extracting a plurality of curb candidate line segments among the line segments on the basis of an angle between the road surface and the line segment; (c) extracting a plurality of curb candidates having a plurality of curb properties, wherein each of the plurality of curb candidates is generated by combining the couple of the curb candidate line segments; and (d) applying the plurality of the curb candidates to a Kernel Fisher Discriminant Analysis to extract a final curb.

10 . The method for localizing of the mobile robot using curb information of the road according to claim 9 , wherein in the step (B), an error for the measurement value of the DGPS module comprises a HDOP (Horizontal Dilution of Precision) error and a Pseudo-range error which is a geometric error according to a geometric constellation of satellites; wherein an error for the measurement value o

Assignees

• Univ Korea Res & Bus Found

Inventors

• Chung Woojin
• Lee Hyunsuk

Classifications

• G01S17/42 Primary

  Simultaneous measurement of distance and other co-ordinates (indirect measurement G01S17/46) · CPC title

• G06F18/2132

  based on discrimination criteria, e.g. discriminant analysis · CPC title

• G01C3/08

  Use of electric radiation detectors · CPC title

• Y10S901/01

  Mobile robot · CPC title

• G01C7/04

  involving a vehicle which moves along the profile to be traced · CPC title