Experimental indoors localization using distance measurements obtained from visual data

In recent years, the emergence of artificial intelligence increases the demand of automatic and robust localization outdoors and indoors. While GPS provides enough accuracy in most outdoor cases, there is still a lack of robust and efficient indoor localization systems available on the market. In this report, an experimental framework for indoor localization is developed and tested. To operate an automatic robot owned by LCAV laboratory, two different operation modes have been successfully implemented, including controlling a robot in real-time or with extra input containing a list of commands. Also, a new visual fiducial system have been developed, and is able to capture the locations of Apriltags inside a room accurately. Multidimensional scaling (MDS) and Squared range least square (SRLS) algorithms containing distance information will also be introduced and the final localization result is within 2% error of tolerance compared with the ground truth results measured by a laser meter.


Advisor(s):
Scheibler, Robin
Dümbgen, Frederike
Vetterli, Martin
Year:
2017
Keywords:
Note:
MASTER_SEMESTER
Laboratories:


Note: The status of this file is: Involved Laboratories Only


 Record created 2017-06-09, last modified 2018-09-13

data results generating during experiments:
Download fulltextZIP
n/a:
Download fulltextPDF
necessary codes for this semester project:
Download fulltextZIP
External link:
Download fulltextURL
Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)