A Robust Feature Matching Method for Robot Localization in a Dynamic Indoor Environment - Technologies and Applications of Artificial Intelligence - page 356

perimeter of geometry of each of the hypothesis point and the perimeter of geometry
of the robot real-time information.

Perimeter of Geometry∑D_Coordinate .

(2)

Laser rangefinder

Start

Local features

Move Tracking

Omni-Directional

Method

Feature Grouping

Area Matching

No

All Hypothesis

Perimeter Matching

Features Grouping

If n<

Overlap Matching

threshold

: The difference between

Yes

the real-time feature and the

Output

hypothetical feature.

Ideal = 0.

Fig. 6. System flowchart

3.3 Overlap Matching

The third feature is the overlap feature. To find which is the most possible direction of
the robot (since the robot does not has a compass), the feature points will be rotated
while matching. Since we have grouped the information of a laser rangefinder into d
groups, the rotation range is defined as