final presentation and live demo seqslam[1]€¦ · seqslam[1] final presentation and live demo tu...
TRANSCRIPT
![Page 1: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/1.jpg)
SeqSLAM[1]
Final Presentation and Live Demo
TU München: Applied Computer Vision for Robotics SS 2013
Tim Wiese, Mathias Kanzler, Maxi Weber
![Page 2: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/2.jpg)
The Problem
Recognition of recurring sceneries under changing environmental
conditions:
● Camera input
● Saving compressed frames
● Applying algorithm to search for suitable matches
● Several improvements to fulfill given requirements
![Page 3: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/3.jpg)
Preprocessing
Store compressed information without loosing relevant information:
● Convert to grayscale
● Scale down to 64 x 36● Contrast enhancement: stretch histogram, patch based
![Page 4: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/4.jpg)
Matching
● Calculate image difference value for
each template
● Local Contrast Enhancement
● Localized Sequence Recognition
Σ
![Page 5: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/5.jpg)
Matching
● Calculate image difference value for
each template
● Local Contrast Enhancement
● Localized Sequence Recognition
![Page 6: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/6.jpg)
Matching
● Calculate image difference value for
each template
● Local Contrast Enhancement
● Localized Sequence Recognition
Image difference matrix
![Page 7: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/7.jpg)
Performance
Decoding andPreprocessing
MatchingGraphicalOutput
100ms
300ms
200ms
![Page 8: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/8.jpg)
Enhancements - Multithreading
Thread 1
Decoding and Preprocessing
Thread 2
Matching
Thread 3
Output
Ring Buffer
![Page 9: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/9.jpg)
Enhancements - Stationary Frame Detection
![Page 10: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/10.jpg)
Results
Successful implementation of SeqSLAM
Robust matching under changing conditions
Multiple times faster than realtime performance
![Page 11: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/11.jpg)
Outlook
● SeqSLAM could be used as independent visual positioning system
● Further improvements by intelligent template learning
○ Learning rate coupled with vehicle speed
○ Online aggregation of subsequent runs with changing environment
![Page 12: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/12.jpg)
ScheduleWeek 1 ● Read paper
● Understand topic● Create schedule
Week 2 ● Start implementation● Preprocess datasets and create bagfiles● Generate first basic output (matches side by side)
Week 3 ● Implementation: final stage● Testing● Bug fixing● Evaluation of matching quality and performance
Week 4 ● Parameter tuning● Improvements (quality / performance)● Enhancements ● Final presentation
Done!Done!Done!
Done!Done!Done!
Done!Done!Done!Done!
Done!Done!Done!
![Page 13: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/13.jpg)
ScheduleWeek 1 ● Read paper
● Understand topic● Create schedule
Week 2 ● Start implementation● Preprocess datasets and create bagfiles● Generate first basic output (matches side by side)
Week 3 ● Implementation: final stage● Testing● Bug fixing● Evaluation of matching quality and performance
Week 4 ● Parameter tuning● Improvements (quality / performance)● Enhancements ● Final presentation
Done!Done!Done!
Done!Done!Done!
Done!Done!Done!Done!
Done!Done!Done!Done!
![Page 14: Final Presentation and Live Demo SeqSLAM[1]€¦ · SeqSLAM[1] Final Presentation and Live Demo TU München: Applied Computer Vision for Robotics SS 2013 Tim Wiese, Mathias Kanzler,](https://reader036.vdocuments.us/reader036/viewer/2022070712/5ecbe150b440392f6b522a51/html5/thumbnails/14.jpg)
...
The End
[1] Michael Milford, Gordon Fraser Wyeth: SeqSLAM: Visual route-based navigation for sunny summer days and stormy winter nights. ICRA 2012: pp. 1643-1649