Difference between revisions of "RTAB-Map"

From IntRoLab
Line 34: Line 34:
 
Figure 2: Processing time for each image acquired (real-time limit fixed to 700 ms for an image rate of 1 Hz)
 
Figure 2: Processing time for each image acquired (real-time limit fixed to 700 ms for an image rate of 1 Hz)
  
Figure 3: Precision-Recall (43% recall at 100% precision)
+
Figure 3: Precision-Recall (48% recall at 100% precision)
  
 
<div style="text-align: center;">
 
<div style="text-align: center;">
Line 55: Line 55:
 
Figure 2: Temps d'exécution pour chaque itération (limite temps réel fixée à 700 ms pour un temps d'acquisition de 1 seconde)
 
Figure 2: Temps d'exécution pour chaque itération (limite temps réel fixée à 700 ms pour un temps d'acquisition de 1 seconde)
  
Figure 3: Precision-Recall (43% recall à 100% precision)
+
Figure 3: Precision-Recall (48% recall à 100% precision)
  
 
<div style="text-align: center;">
 
<div style="text-align: center;">
Line 61: Line 61:
 
</div>
 
</div>
  
 +
'''Vidéos'''
 +
<code>{{#ev:youtube|ShQlakkzsY4}} {{#ev:youtube|cTmf5yrpcl8}}</code>
 +
<code>{{#ev:youtube|CuWESlLfWpQ}} {{#ev:youtube|SQiFs1z7qSY}}</code>
 
</french>
 
</french>
  
Line 84: Line 87:
  
 
* 5395 images at 1 Hz (1.5 hours).
 
* 5395 images at 1 Hz (1.5 hours).
* Walking the same loop of ~2 km two times.
+
* Images taken while walking through a loop of ~2 km, traversed two times.
 
* The data set contains indoor and outdoor environments.
 
* The data set contains indoor and outdoor environments.
 
<div style="text-align: center;">
 
<div style="text-align: center;">
Line 155: Line 158:
 
<french>
 
<french>
 
== Ensembles de données ==
 
== Ensembles de données ==
====UdeS====
+
'''UdeS'''
 +
 
 +
* 5395 images à 1 Hz (1,5 heures).
 +
* Images prises en marchant sur un trajet de ~2 km, parcouru deux fois.
 +
* L'ensemble de données contient des images prises à l'intérieur et à l'extérieur.
 +
<div style="text-align: center;">
 +
[[File:UdeS_1Hz_map.png|250px]]
 +
([http://maps.google.ca/maps?q=Universit%C3%A9+de+sherbrooke&hl=en&ie=UTF8&ll=45.377714,-71.927383&spn=0.011546,0.016158&sll=49.891235,-97.15369&sspn=43.664668,66.181641&t=h&z=16 sur Google maps])
 +
</div>
 
<gallery perrow=5>
 
<gallery perrow=5>
 
File:UdeS_1Hz_sample1.jpg
 
File:UdeS_1Hz_sample1.jpg
Line 173: Line 184:
 
File:UdeS_1Hz_sample15.jpg
 
File:UdeS_1Hz_sample15.jpg
 
</gallery>
 
</gallery>
====Téléchargements====
+
  [[Media:UdeS_1Hz.part1.rar|UdeS_1Hz.part1.rar]]
Ensemble de données :
+
  [[Media:UdeS_1Hz.part2.rar|UdeS_1Hz.part2.rar]]
* UdeS (5395 images à 1 Hz)
+
  [[Media:UdeS_1Hz.part3.rar|UdeS_1Hz.part3.rar]]
  * [[Media:UdeS_1Hz.part1.rar|UdeS_1Hz.part1.rar]]
+
  [[Media:UdeS_1Hz.rar|UdeS_1Hz GroundTruth]]
  * [[Media:UdeS_1Hz.part2.rar|UdeS_1Hz.part2.rar]]
+
 
  * [[Media:UdeS_1Hz.part3.rar|UdeS_1Hz.part3.rar]]
+
'''NFSMW'''
  * [[Media:UdeS_1Hz.rar|UdeS_1Hz GroundTruth]]
+
 
* NFSMW (25098 images à 1 Hz)
+
* Images prises dans le jeu vidéo de course Need For Speed: Most Wanted.
  * [[Media:NFSMW_1Hz.part01.rar|NFSMW_1Hz.part01.rar]]
+
* 2 zones ont été visités 100 fois chaque (100 boucles dans la zone 1 et ensuite 102 boucles dans la zone 2).
  * [[Media:NFSMW_1Hz.part02.rar|NFSMW_1Hz.part02.rar]]
+
* 25098 images à 1 Hz (7 heures).
  * [[Media:NFSMW_1Hz.part03.rar|NFSMW_1Hz.part03.rar]]
+
<div style="text-align: center;">
  * [[Media:NFSMW_1Hz.part04.rar|NFSMW_1Hz.part04.rar]]
+
[[File:NFSMW_1Hz_map.png|250px]]
  * [[Media:NFSMW_1Hz.part05.rar|NFSMW_1Hz.part05.rar]]
+
</div>
  * [[Media:NFSMW_1Hz.part06.rar|NFSMW_1Hz.part06.rar]]
+
<gallery perrow=5>
  * [[Media:NFSMW_1Hz.part07.rar|NFSMW_1Hz.part07.rar]]
+
File:NFSMW_1Hz_sample1.jpg
  * [[Media:NFSMW_1Hz.part08.rar|NFSMW_1Hz.part08.rar]]
+
File:NFSMW_1Hz_sample2.jpg
 +
File:NFSMW_1Hz_sample3.jpg
 +
File:NFSMW_1Hz_sample4.jpg
 +
File:NFSMW_1Hz_sample5.jpg
 +
File:NFSMW_1Hz_sample6.jpg|Comparer l'illumination avec l'image suivante...
 +
File:NFSMW_1Hz_sample8.jpg
 +
File:NFSMW_1Hz_sample7.jpg
 +
File:NFSMW_1Hz_sample9.jpg
 +
File:NFSMW_1Hz_sample10.jpg
 +
</gallery>
 +
  [[Media:NFSMW_1Hz.part01.rar|NFSMW_1Hz.part01.rar]]
 +
  [[Media:NFSMW_1Hz.part02.rar|NFSMW_1Hz.part02.rar]]
 +
  [[Media:NFSMW_1Hz.part03.rar|NFSMW_1Hz.part03.rar]]
 +
  [[Media:NFSMW_1Hz.part04.rar|NFSMW_1Hz.part04.rar]]
 +
  [[Media:NFSMW_1Hz.part05.rar|NFSMW_1Hz.part05.rar]]
 +
  [[Media:NFSMW_1Hz.part06.rar|NFSMW_1Hz.part06.rar]]
 +
  [[Media:NFSMW_1Hz.part07.rar|NFSMW_1Hz.part07.rar]]
 +
  [[Media:NFSMW_1Hz.part08.rar|NFSMW_1Hz.part08.rar]]
 +
 
 +
'''Communauté'''
  
 
Ensembles de données provenant d'autres approches de détection de fermeture de boucle :
 
Ensembles de données provenant d'autres approches de détection de fermeture de boucle :
Line 195: Line 225:
  
 
Ground truths (format lisible par RTAB-Map) :
 
Ground truths (format lisible par RTAB-Map) :
* [[Media:NewCollege.rar|NewCollege.rar]] (with the left and right images merged)
+
* [[Media:NewCollege.rar|NewCollege.rar]] 1073 images à ~0.5 Hz (les images de gauche et de droite fusionnées)
* [[Media:CityCentre.rar|CityCentre.rar]] (with the left and right images merged)
+
* [[Media:CityCentre.rar|CityCentre.rar]] 1237 images à ~0.5 Hz (les images de gauche et de droite fusionnées)
* [[Media:Lip6Indoor.rar|Lip6Indoor.rar]]
+
* [[Media:Lip6Indoor.rar|Lip6Indoor.rar]] 388 images à 1 Hz
* [[Media:Lip6Outdoor.rar|Lip6Outdoor.rar]]
+
* [[Media:Lip6Outdoor.rar|Lip6Outdoor.rar]] 531 images à 0.5 Hz
 
</french>
 
</french>
  

Revision as of 20:41, 25 July 2011

RTAB-Map : Real-Time Appearance-Based Mapping

Description[edit]

Loop closure detection is the process involved when trying to find a match between the current and a previously visited locations in SLAM (Simultaneous Localization And Mapping). Over time, the amount of time required to process new observations increases with the size of the internal map, which may affect real-time processing. RTAB-Map is a novel real-time loop closure detection approach for large-scale and long-term SLAM. Our approach is based on efficient memory management to keep computation time for each new observation under a fixed time limit, thus respecting real-time limit for long-term operation. Results demonstrate the approach's adaptability and scalability using one custom data set and four standard data sets.

Example of sensorimotor learning using directly this loop closure detection approach (new in RTAB-Map 0.3) :

Results[edit]

Note that these results (more recent) may differ from those in the video...

Figure 1: Summary of the loop closures detected on UdeS data set :

  • Green : Loop closures detected
  • Yellow : Loop closures rejected
  • Red : Unable to detect a loop closure because old places could not be retrieved

Figure 2: Processing time for each image acquired (real-time limit fixed to 700 ms for an image rate of 1 Hz)

Figure 3: Precision-Recall (48% recall at 100% precision)

RTAB-Map LoopClosureMapResults.png RTAB-Map LoopClosureTimeResults.png RTAB-Map RecallResults.png

Videos


Source code[edit]

The code was tested on Windows (Xp, 7), Mac OS X 10.6 and Ubuntu 10.4LTS.

Images acquired in Need For Speed Most Wanted

Data sets[edit]

UdeS

  • 5395 images at 1 Hz (1.5 hours).
  • Images taken while walking through a loop of ~2 km, traversed two times.
  • The data set contains indoor and outdoor environments.
UdeS_1Hz.part1.rar
UdeS_1Hz.part2.rar
UdeS_1Hz.part3.rar
UdeS_1Hz GroundTruth

NFSMW

  • Images taken from the racing video game Need For Speed: Most Wanted.
  • 2 areas visited hundred times each (100 traversals in area 1 then moved to area 2 for another 102 traversals).
  • 25098 images at 1 Hz (7 hours).

NFSMW 1Hz map.png

NFSMW_1Hz.part01.rar
NFSMW_1Hz.part02.rar
NFSMW_1Hz.part03.rar
NFSMW_1Hz.part04.rar
NFSMW_1Hz.part05.rar
NFSMW_1Hz.part06.rar
NFSMW_1Hz.part07.rar
NFSMW_1Hz.part08.rar

Community

Community data sets from other loop closure detection approaches :

Ground truths (readable by RTAB-Map) :


Publications

M. Labbé and F. Michaud, “Memory management for real-time appearance-based loop closure detection,” in Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems, 2011. (Accepted)