Information Technology Reference
In-Depth Information
Fig. 5. Divide the information into 8 groups
3
Feature Matching and Move Tracking
The system flowchart is shown in Fig. 6. The robot use a laser rangefinder to gather
local features. After grouping the features into 8 groups, our system performs the
feature matching process. Our feature matching method is an enhanced version of the
signature-rs method in a previous work [9]. The previous method contains three steps
of matching to filter out impossible location candidates and to reduce the candidates
that the system needs to do the feature matching. Our system applies the same steps to
the 8 groups of features, and preserved the matched cases. If the number of matched
cases excesses the threshold, the location is found. Otherwise, a move tracking
method is necessary to guide the robot to move. The feature matching and move
tracking are as follows.
3.1
Area Matching
For each given hypothetical feature point (HFP), the length of each feature is stored in
the Feature Array (F_Array). Adding up the length in the F_Array can be a way to
calculate the area (Area of Geometry)
The first feature matching is actually a
comparison between the area of geometry of each of the hypothesis point and the area
of geometry of the robot real-time information.
(1)
Area of Geometry∑F_Array .
(1)
3.2
Perimeter Matching
The second feature is called the perimeter feature. We translate the information in the
feature array (F_Array) as the location point on a plane, which is called depth
coordinate (D_Coordinate). By connecting all the points in D_Coordinate, we get the
perimeter (Perimeter of Geometry) (2). The second feature matching is comparing the
Search WWH ::




Custom Search