國立虎尾科技大學 |

Multi-Agent Visual Slam.

紀錄類型:	書目-語言資料,手稿 : Monograph/item
正題名/作者:	Multi-Agent Visual Slam./
作者:	Andresen, Wyatt Cole.
面頁冊數:	1 online resource (50 pages)
附註:	Source: Masters Abstracts International, Volume: 84-04.
Contained By:	Masters Abstracts International84-04.
標題:	Computer science. -
電子資源:	click for full text (PQDT)
ISBN:	9798351499871

Multi-Agent Visual Slam.
Andresen, Wyatt Cole.

Multi-Agent Visual Slam. - 1 online resource (50 pages)

Source: Masters Abstracts International, Volume: 84-04.

Thesis (M.Sc.)--North Carolina State University, 2022.

Includes bibliographical references

In visual simultaneous localization and mapping (vSLAM), camera(s) attached to an agent are used to construct a map of the environment and determine the location of the agent within the environment. As vSLAM algorithms have matured in recent years, multiagent vSLAM, where multiple agents collaboratively perform SLAM together, has become an important area of research. Much of this research focuses on the core architectural and communications problems that are inherent to centralized and decentralized multi-agent systems of all kinds. More exclusive to multi-agent vSLAM, however, is the problem of combining the maps of multiple agents, called map fusion. In this thesis a real-time system for performing multi-agent vSLAM is proposed. Specifically, a fully centralized multi-agent architecture is implemented to minimize the communications issues in centralized and decentralized architectures. Then, a map fusion algorithm is proposed that efficiently merges two maps into one using loop closure techniques such as place recognition. Novelly, covisibility connections are identified globally between the maps, not only in the area of overlap, in a step called covisibility discovery. The performance of the proposed system is evaluated qualitatively against ground truth trajectories by visual inspection, then quantitatively compared to ORB-SLAM2 with absolute and relative pose errors. The results show no to low loss of accuracy, excellent global discovery of covisibility connections, and near real-time performance within limited computational resources.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798351499871Subjects--Topical Terms:

573171
Computer science.
Index Terms--Genre/Form:

554714
Electronic books.

Multi-Agent Visual Slam.
LDR:02765ntm a22003497 4500 001 1144234
005 20240531083806.5
006 m o d
007 cr mn ---uuuuu
008 250605s2022 xx obm 000 0 eng d
020 $a 9798351499871
035 $a (MiAaPQ)AAI29342792
035 $a (MiAaPQ)NCState_Univ18402039722
035 $a AAI29342792
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Andresen, Wyatt Cole. $3 1469196
245 1 0 $a Multi-Agent Visual Slam.
264 0 $c 2022
300 $a 1 online resource (50 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Masters Abstracts International, Volume: 84-04.
500 $a Advisor: Han, Kevin;Ore, John-Paul.
502 $a Thesis (M.Sc.)--North Carolina State University, 2022.
504 $a Includes bibliographical references
520 $a In visual simultaneous localization and mapping (vSLAM), camera(s) attached to an agent are used to construct a map of the environment and determine the location of the agent within the environment. As vSLAM algorithms have matured in recent years, multiagent vSLAM, where multiple agents collaboratively perform SLAM together, has become an important area of research. Much of this research focuses on the core architectural and communications problems that are inherent to centralized and decentralized multi-agent systems of all kinds. More exclusive to multi-agent vSLAM, however, is the problem of combining the maps of multiple agents, called map fusion. In this thesis a real-time system for performing multi-agent vSLAM is proposed. Specifically, a fully centralized multi-agent architecture is implemented to minimize the communications issues in centralized and decentralized architectures. Then, a map fusion algorithm is proposed that efficiently merges two maps into one using loop closure techniques such as place recognition. Novelly, covisibility connections are identified globally between the maps, not only in the area of overlap, in a step called covisibility discovery. The performance of the proposed system is evaluated qualitatively against ground truth trajectories by visual inspection, then quantitatively compared to ORB-SLAM2 with absolute and relative pose errors. The results show no to low loss of accuracy, excellent global discovery of covisibility connections, and near real-time performance within limited computational resources.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Computer science. $3 573171
650 4 $a Robotics. $3 561941
650 4 $a Sensors. $3 1003702
650 4 $a Optimization. $3 669174
650 4 $a Localization. $3 1091696
650 4 $a Algorithms. $3 527865
650 4 $a Usability. $3 1062032
650 4 $a Cameras. $3 812890
650 4 $a Mapping. $3 1215615
650 4 $a Maps. $3 889832
655 7 $a Electronic books. $2 local $3 554714
690 $a 0771
690 $a 0729
690 $a 0984
710 2 $a North Carolina State University. $3 845424
710 2 $a ProQuest Information and Learning Co. $3 1178819
773 0 $t Masters Abstracts International $g 84-04.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29342792 $z click for full text (PQDT)