SlideShare ist ein Scribd-Unternehmen logo
1 von 34
1 / 32
All roads lead to ROMA
Friday, 6 March 2014
Roelof P. van den Berg
2 / 32Roelof P. van den Berg
Setting the scene
• disaster relief scenario
• no up-to-date map available
• changing environment
• requirements
• fast construction of a best guess
• adapt to new information
3 / 32Roelof P. van den Berg
Outline
1. Introduction
• Research question
• Relevance
• Methodology
2. Algorithm
3. Experiments
4. Discussion
5. Conclusion
4 / 32Roelof P. van den Berg
Introduction – Research Question
How to create and maintain
an accurate and dynamic map
based on position traces?
5 / 32Roelof P. van den Berg
Introduction – Relevance
• scientific
• insight in map dynamics
• novel description of road decay
• novel method of map comparison
• practical
• situations with no a priori map
• speedy creation of a best guess map
6 / 32Roelof P. van den Berg
Introduction – Methodology 1
3
7 / 32Roelof P. van den Berg
Introduction – Methodology 2
3
8 / 32Roelof P. van den Berg
Introduction – Methodology 3
3
Zoeterwoude-Dorp
Inhabitants:
• 4.514
Traffic model:
• shortest path
• random
9 / 32Roelof P. van den Berg
Outline
1. Introduction
2. Algorithm
3. Experiments
4. Discussion
5. Conclusion
10 / 32Roelof P. van den Berg
Algorithm – Layout
I. Pre-processing
II. Path estimation
III. Path generation
IV. Path adjustment
V. Map dynamics node
edge
measurement
11 / 32Roelof P. van den Berg
Algorithm – Pre-processing
I. Pre-processing
• Filter on accuracy
• Filter on distance
II. Path estimation
III. Path generation
IV. Path adjustment
V. Map dynamics
12 / 32Roelof P. van den Berg
Algorithm – Path estimation
I. Pre-processing
II. Path estimation
• Match edges to measurements
• Apply forward tree search
III. Path generation
IV. Path adjustment
V. Map dynamics
13 / 32Roelof P. van den Berg
Algorithm – Path generation 1
2
I. Pre-processing
II. Path estimation
III.Path generation
IV. Path adjustment
V. Map dynamics
14 / 32Roelof P. van den Berg
Algorithm – Path generation 2
2
I. Pre-processing
II. Path estimation
III.Path generation
• Backward tree search
• Bridge shortest gap
IV. Path adjustment
V. Map dynamics
15 / 32Roelof P. van den Berg
Algorithm – Path adjustment
I. Pre-processing
II. Path estimation
III. Path generation
IV.Path adjustment
• measurement interpolation
• node influencing
V. Map dynamics
16 / 32Roelof P. van den Berg
Algorithm – Map dynamics
I. Pre-processing
II. Path estimation
III. Path generation
IV. Path adjustment
V. Map dynamics
• node merging
• road decay
average interval
time
position
PP CDF-1
17 / 32Roelof P. van den Berg
Outline
1. Introduction
2. Algorithm
3. Experiments
• Comparison
• Road introduction
• Road removal
• Map dynamics
4. Discussion
5. Conclusion
18 / 32Roelof P. van den Berg
Two vector-maps
Map & World .
Match nodes
Map <-> World
Experiments – Comparison 1
2
19 / 32Roelof P. van den Berg
Experiments – Comparison 2
2
Recall =
𝑖=0
|𝑆|
𝑗=0
|𝑆|
𝑆 𝑖,𝑗 𝑊 𝑖,𝑗
|𝑒 𝑤|
Precision =
𝑖=0
|𝑆|
𝑗=0
|𝑆|
𝑆 𝑖,𝑗 𝑀 𝑖,𝑗
|𝑒 𝑚|
F-score = 2 ×
prec × rec
prec + rec
False Positive
False Negative
True Positive
Similarity
20 / 32Roelof P. van den Berg
Experiments – Road Introduction 1
2
• Map stabilizes
• for 50, 100, and 200 cars
• Clear overfitting
• for ≥ 1000 cars simulated
• within 3.5 – 6.5 minutes
• faster for more traffic
21 / 32Roelof P. van den Berg
Experiments – Road Introduction 2
2
observed overfitting
2000 cars; overfitting at 5.5 minutes
22 / 32Roelof P. van den Berg
Experiments – Road removal 1
2
• Roadblocks added
• 10, 20, and 50%
• Map recovers
• generally within 30 minutes
• faster for more traffic
23 / 32Roelof P. van den Berg
Experiments – Road removal 2
2
road removal
recovery
2000 cars; recovery within 25 minutes
24 / 32Roelof P. van den Berg
Road introduction Road removal
Experiments – Map dynamics
25 / 32Roelof P. van den Berg
Outline
1. Introduction
2. Algorithm
3. Experiments
4. Discussion
• Novelties
• Noise
• Improvements
5. Conclusion
26 / 32Roelof P. van den Berg
Discussion – Novelties
• Description of road decay method
• Investigation in formation of noise
• Graph based method for map comparison
27 / 32Roelof P. van den Berg
Discussion – Noise
• Outliers in measurements
• Many roads on the map
• One road in the world
• Creep towards centreline
28 / 32Roelof P. van den Berg
Discussion – Improvements
• Traffic model in simulator
• Filtering methods
• Outlier removal
• Road decay
29 / 32Roelof P. van den Berg
Outline
1. Introduction
2. Algorithm
3. Experiments
4. Discussion
5. Conclusion
• Conclusion
• Future work
“How to create and maintain
an accurate and dynamic map
based on position traces?”
30 / 32Roelof P. van den Berg
Conclusion
• ROMA is a proof of concept
• Dynamic vector-map generation in an online fashion
• Framework for map-generation evaluation
31 / 32Roelof P. van den Berg
Conclusion – Future work
• Topological and geographical map dynamics
• Distributed application of ROMA
Alice knows:
Bob knows:
32 / 32
All roads lead to ROMA
Friday, 6 March 2014
Roelof P. van den Berg
33 / 32Roelof P. van den Berg
Appendix – Road Introduction
34 / 32Roelof P. van den Berg
Appendix – Road Removal

Weitere ähnliche Inhalte

Kürzlich hochgeladen

The basics of sentences session 2pptx copy.pptx
The basics of sentences session 2pptx copy.pptxThe basics of sentences session 2pptx copy.pptx
The basics of sentences session 2pptx copy.pptxheathfieldcps1
 
Class 11th Physics NEET formula sheet pdf
Class 11th Physics NEET formula sheet pdfClass 11th Physics NEET formula sheet pdf
Class 11th Physics NEET formula sheet pdfAyushMahapatra5
 
Sanyam Choudhary Chemistry practical.pdf
Sanyam Choudhary Chemistry practical.pdfSanyam Choudhary Chemistry practical.pdf
Sanyam Choudhary Chemistry practical.pdfsanyamsingh5019
 
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in DelhiRussian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhikauryashika82
 
BASLIQ CURRENT LOOKBOOK LOOKBOOK(1) (1).pdf
BASLIQ CURRENT LOOKBOOK  LOOKBOOK(1) (1).pdfBASLIQ CURRENT LOOKBOOK  LOOKBOOK(1) (1).pdf
BASLIQ CURRENT LOOKBOOK LOOKBOOK(1) (1).pdfSoniaTolstoy
 
Beyond the EU: DORA and NIS 2 Directive's Global Impact
Beyond the EU: DORA and NIS 2 Directive's Global ImpactBeyond the EU: DORA and NIS 2 Directive's Global Impact
Beyond the EU: DORA and NIS 2 Directive's Global ImpactPECB
 
Z Score,T Score, Percential Rank and Box Plot Graph
Z Score,T Score, Percential Rank and Box Plot GraphZ Score,T Score, Percential Rank and Box Plot Graph
Z Score,T Score, Percential Rank and Box Plot GraphThiyagu K
 
Arihant handbook biology for class 11 .pdf
Arihant handbook biology for class 11 .pdfArihant handbook biology for class 11 .pdf
Arihant handbook biology for class 11 .pdfchloefrazer622
 
Paris 2024 Olympic Geographies - an activity
Paris 2024 Olympic Geographies - an activityParis 2024 Olympic Geographies - an activity
Paris 2024 Olympic Geographies - an activityGeoBlogs
 
Software Engineering Methodologies (overview)
Software Engineering Methodologies (overview)Software Engineering Methodologies (overview)
Software Engineering Methodologies (overview)eniolaolutunde
 
Call Girls in Dwarka Mor Delhi Contact Us 9654467111
Call Girls in Dwarka Mor Delhi Contact Us 9654467111Call Girls in Dwarka Mor Delhi Contact Us 9654467111
Call Girls in Dwarka Mor Delhi Contact Us 9654467111Sapana Sha
 
A Critique of the Proposed National Education Policy Reform
A Critique of the Proposed National Education Policy ReformA Critique of the Proposed National Education Policy Reform
A Critique of the Proposed National Education Policy ReformChameera Dedduwage
 
Student login on Anyboli platform.helpin
Student login on Anyboli platform.helpinStudent login on Anyboli platform.helpin
Student login on Anyboli platform.helpinRaunakKeshri1
 
The Most Excellent Way | 1 Corinthians 13
The Most Excellent Way | 1 Corinthians 13The Most Excellent Way | 1 Corinthians 13
The Most Excellent Way | 1 Corinthians 13Steve Thomason
 
Grant Readiness 101 TechSoup and Remy Consulting
Grant Readiness 101 TechSoup and Remy ConsultingGrant Readiness 101 TechSoup and Remy Consulting
Grant Readiness 101 TechSoup and Remy ConsultingTechSoup
 
Measures of Central Tendency: Mean, Median and Mode
Measures of Central Tendency: Mean, Median and ModeMeasures of Central Tendency: Mean, Median and Mode
Measures of Central Tendency: Mean, Median and ModeThiyagu K
 
SOCIAL AND HISTORICAL CONTEXT - LFTVD.pptx
SOCIAL AND HISTORICAL CONTEXT - LFTVD.pptxSOCIAL AND HISTORICAL CONTEXT - LFTVD.pptx
SOCIAL AND HISTORICAL CONTEXT - LFTVD.pptxiammrhaywood
 

Kürzlich hochgeladen (20)

The basics of sentences session 2pptx copy.pptx
The basics of sentences session 2pptx copy.pptxThe basics of sentences session 2pptx copy.pptx
The basics of sentences session 2pptx copy.pptx
 
Class 11th Physics NEET formula sheet pdf
Class 11th Physics NEET formula sheet pdfClass 11th Physics NEET formula sheet pdf
Class 11th Physics NEET formula sheet pdf
 
Sanyam Choudhary Chemistry practical.pdf
Sanyam Choudhary Chemistry practical.pdfSanyam Choudhary Chemistry practical.pdf
Sanyam Choudhary Chemistry practical.pdf
 
Mattingly "AI & Prompt Design: The Basics of Prompt Design"
Mattingly "AI & Prompt Design: The Basics of Prompt Design"Mattingly "AI & Prompt Design: The Basics of Prompt Design"
Mattingly "AI & Prompt Design: The Basics of Prompt Design"
 
Advance Mobile Application Development class 07
Advance Mobile Application Development class 07Advance Mobile Application Development class 07
Advance Mobile Application Development class 07
 
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in DelhiRussian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
 
BASLIQ CURRENT LOOKBOOK LOOKBOOK(1) (1).pdf
BASLIQ CURRENT LOOKBOOK  LOOKBOOK(1) (1).pdfBASLIQ CURRENT LOOKBOOK  LOOKBOOK(1) (1).pdf
BASLIQ CURRENT LOOKBOOK LOOKBOOK(1) (1).pdf
 
Beyond the EU: DORA and NIS 2 Directive's Global Impact
Beyond the EU: DORA and NIS 2 Directive's Global ImpactBeyond the EU: DORA and NIS 2 Directive's Global Impact
Beyond the EU: DORA and NIS 2 Directive's Global Impact
 
Z Score,T Score, Percential Rank and Box Plot Graph
Z Score,T Score, Percential Rank and Box Plot GraphZ Score,T Score, Percential Rank and Box Plot Graph
Z Score,T Score, Percential Rank and Box Plot Graph
 
Arihant handbook biology for class 11 .pdf
Arihant handbook biology for class 11 .pdfArihant handbook biology for class 11 .pdf
Arihant handbook biology for class 11 .pdf
 
Paris 2024 Olympic Geographies - an activity
Paris 2024 Olympic Geographies - an activityParis 2024 Olympic Geographies - an activity
Paris 2024 Olympic Geographies - an activity
 
Software Engineering Methodologies (overview)
Software Engineering Methodologies (overview)Software Engineering Methodologies (overview)
Software Engineering Methodologies (overview)
 
Call Girls in Dwarka Mor Delhi Contact Us 9654467111
Call Girls in Dwarka Mor Delhi Contact Us 9654467111Call Girls in Dwarka Mor Delhi Contact Us 9654467111
Call Girls in Dwarka Mor Delhi Contact Us 9654467111
 
A Critique of the Proposed National Education Policy Reform
A Critique of the Proposed National Education Policy ReformA Critique of the Proposed National Education Policy Reform
A Critique of the Proposed National Education Policy Reform
 
Student login on Anyboli platform.helpin
Student login on Anyboli platform.helpinStudent login on Anyboli platform.helpin
Student login on Anyboli platform.helpin
 
The Most Excellent Way | 1 Corinthians 13
The Most Excellent Way | 1 Corinthians 13The Most Excellent Way | 1 Corinthians 13
The Most Excellent Way | 1 Corinthians 13
 
Grant Readiness 101 TechSoup and Remy Consulting
Grant Readiness 101 TechSoup and Remy ConsultingGrant Readiness 101 TechSoup and Remy Consulting
Grant Readiness 101 TechSoup and Remy Consulting
 
Measures of Central Tendency: Mean, Median and Mode
Measures of Central Tendency: Mean, Median and ModeMeasures of Central Tendency: Mean, Median and Mode
Measures of Central Tendency: Mean, Median and Mode
 
INDIA QUIZ 2024 RLAC DELHI UNIVERSITY.pptx
INDIA QUIZ 2024 RLAC DELHI UNIVERSITY.pptxINDIA QUIZ 2024 RLAC DELHI UNIVERSITY.pptx
INDIA QUIZ 2024 RLAC DELHI UNIVERSITY.pptx
 
SOCIAL AND HISTORICAL CONTEXT - LFTVD.pptx
SOCIAL AND HISTORICAL CONTEXT - LFTVD.pptxSOCIAL AND HISTORICAL CONTEXT - LFTVD.pptx
SOCIAL AND HISTORICAL CONTEXT - LFTVD.pptx
 

Empfohlen

2024 State of Marketing Report – by Hubspot
2024 State of Marketing Report – by Hubspot2024 State of Marketing Report – by Hubspot
2024 State of Marketing Report – by HubspotMarius Sescu
 
Everything You Need To Know About ChatGPT
Everything You Need To Know About ChatGPTEverything You Need To Know About ChatGPT
Everything You Need To Know About ChatGPTExpeed Software
 
Product Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage EngineeringsProduct Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage EngineeringsPixeldarts
 
How Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental HealthHow Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental HealthThinkNow
 
AI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfAI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfmarketingartwork
 
PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024Neil Kimberley
 
Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)contently
 
How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024Albert Qian
 
Social Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsSocial Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsKurio // The Social Media Age(ncy)
 
Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Search Engine Journal
 
5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summarySpeakerHub
 
ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd Clark Boyd
 
Getting into the tech field. what next
Getting into the tech field. what next Getting into the tech field. what next
Getting into the tech field. what next Tessa Mero
 
Google's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentGoogle's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentLily Ray
 
Time Management & Productivity - Best Practices
Time Management & Productivity -  Best PracticesTime Management & Productivity -  Best Practices
Time Management & Productivity - Best PracticesVit Horky
 
The six step guide to practical project management
The six step guide to practical project managementThe six step guide to practical project management
The six step guide to practical project managementMindGenius
 
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...RachelPearson36
 

Empfohlen (20)

2024 State of Marketing Report – by Hubspot
2024 State of Marketing Report – by Hubspot2024 State of Marketing Report – by Hubspot
2024 State of Marketing Report – by Hubspot
 
Everything You Need To Know About ChatGPT
Everything You Need To Know About ChatGPTEverything You Need To Know About ChatGPT
Everything You Need To Know About ChatGPT
 
Product Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage EngineeringsProduct Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage Engineerings
 
How Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental HealthHow Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental Health
 
AI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfAI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdf
 
Skeleton Culture Code
Skeleton Culture CodeSkeleton Culture Code
Skeleton Culture Code
 
PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024
 
Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)
 
How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024
 
Social Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsSocial Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie Insights
 
Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024
 
5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary
 
ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd
 
Getting into the tech field. what next
Getting into the tech field. what next Getting into the tech field. what next
Getting into the tech field. what next
 
Google's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentGoogle's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search Intent
 
How to have difficult conversations
How to have difficult conversations How to have difficult conversations
How to have difficult conversations
 
Introduction to Data Science
Introduction to Data ScienceIntroduction to Data Science
Introduction to Data Science
 
Time Management & Productivity - Best Practices
Time Management & Productivity -  Best PracticesTime Management & Productivity -  Best Practices
Time Management & Productivity - Best Practices
 
The six step guide to practical project management
The six step guide to practical project managementThe six step guide to practical project management
The six step guide to practical project management
 
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
 

All roads lead to ROMA

  • 1. 1 / 32 All roads lead to ROMA Friday, 6 March 2014 Roelof P. van den Berg
  • 2. 2 / 32Roelof P. van den Berg Setting the scene • disaster relief scenario • no up-to-date map available • changing environment • requirements • fast construction of a best guess • adapt to new information
  • 3. 3 / 32Roelof P. van den Berg Outline 1. Introduction • Research question • Relevance • Methodology 2. Algorithm 3. Experiments 4. Discussion 5. Conclusion
  • 4. 4 / 32Roelof P. van den Berg Introduction – Research Question How to create and maintain an accurate and dynamic map based on position traces?
  • 5. 5 / 32Roelof P. van den Berg Introduction – Relevance • scientific • insight in map dynamics • novel description of road decay • novel method of map comparison • practical • situations with no a priori map • speedy creation of a best guess map
  • 6. 6 / 32Roelof P. van den Berg Introduction – Methodology 1 3
  • 7. 7 / 32Roelof P. van den Berg Introduction – Methodology 2 3
  • 8. 8 / 32Roelof P. van den Berg Introduction – Methodology 3 3 Zoeterwoude-Dorp Inhabitants: • 4.514 Traffic model: • shortest path • random
  • 9. 9 / 32Roelof P. van den Berg Outline 1. Introduction 2. Algorithm 3. Experiments 4. Discussion 5. Conclusion
  • 10. 10 / 32Roelof P. van den Berg Algorithm – Layout I. Pre-processing II. Path estimation III. Path generation IV. Path adjustment V. Map dynamics node edge measurement
  • 11. 11 / 32Roelof P. van den Berg Algorithm – Pre-processing I. Pre-processing • Filter on accuracy • Filter on distance II. Path estimation III. Path generation IV. Path adjustment V. Map dynamics
  • 12. 12 / 32Roelof P. van den Berg Algorithm – Path estimation I. Pre-processing II. Path estimation • Match edges to measurements • Apply forward tree search III. Path generation IV. Path adjustment V. Map dynamics
  • 13. 13 / 32Roelof P. van den Berg Algorithm – Path generation 1 2 I. Pre-processing II. Path estimation III.Path generation IV. Path adjustment V. Map dynamics
  • 14. 14 / 32Roelof P. van den Berg Algorithm – Path generation 2 2 I. Pre-processing II. Path estimation III.Path generation • Backward tree search • Bridge shortest gap IV. Path adjustment V. Map dynamics
  • 15. 15 / 32Roelof P. van den Berg Algorithm – Path adjustment I. Pre-processing II. Path estimation III. Path generation IV.Path adjustment • measurement interpolation • node influencing V. Map dynamics
  • 16. 16 / 32Roelof P. van den Berg Algorithm – Map dynamics I. Pre-processing II. Path estimation III. Path generation IV. Path adjustment V. Map dynamics • node merging • road decay average interval time position PP CDF-1
  • 17. 17 / 32Roelof P. van den Berg Outline 1. Introduction 2. Algorithm 3. Experiments • Comparison • Road introduction • Road removal • Map dynamics 4. Discussion 5. Conclusion
  • 18. 18 / 32Roelof P. van den Berg Two vector-maps Map & World . Match nodes Map <-> World Experiments – Comparison 1 2
  • 19. 19 / 32Roelof P. van den Berg Experiments – Comparison 2 2 Recall = 𝑖=0 |𝑆| 𝑗=0 |𝑆| 𝑆 𝑖,𝑗 𝑊 𝑖,𝑗 |𝑒 𝑤| Precision = 𝑖=0 |𝑆| 𝑗=0 |𝑆| 𝑆 𝑖,𝑗 𝑀 𝑖,𝑗 |𝑒 𝑚| F-score = 2 × prec × rec prec + rec False Positive False Negative True Positive Similarity
  • 20. 20 / 32Roelof P. van den Berg Experiments – Road Introduction 1 2 • Map stabilizes • for 50, 100, and 200 cars • Clear overfitting • for ≥ 1000 cars simulated • within 3.5 – 6.5 minutes • faster for more traffic
  • 21. 21 / 32Roelof P. van den Berg Experiments – Road Introduction 2 2 observed overfitting 2000 cars; overfitting at 5.5 minutes
  • 22. 22 / 32Roelof P. van den Berg Experiments – Road removal 1 2 • Roadblocks added • 10, 20, and 50% • Map recovers • generally within 30 minutes • faster for more traffic
  • 23. 23 / 32Roelof P. van den Berg Experiments – Road removal 2 2 road removal recovery 2000 cars; recovery within 25 minutes
  • 24. 24 / 32Roelof P. van den Berg Road introduction Road removal Experiments – Map dynamics
  • 25. 25 / 32Roelof P. van den Berg Outline 1. Introduction 2. Algorithm 3. Experiments 4. Discussion • Novelties • Noise • Improvements 5. Conclusion
  • 26. 26 / 32Roelof P. van den Berg Discussion – Novelties • Description of road decay method • Investigation in formation of noise • Graph based method for map comparison
  • 27. 27 / 32Roelof P. van den Berg Discussion – Noise • Outliers in measurements • Many roads on the map • One road in the world • Creep towards centreline
  • 28. 28 / 32Roelof P. van den Berg Discussion – Improvements • Traffic model in simulator • Filtering methods • Outlier removal • Road decay
  • 29. 29 / 32Roelof P. van den Berg Outline 1. Introduction 2. Algorithm 3. Experiments 4. Discussion 5. Conclusion • Conclusion • Future work “How to create and maintain an accurate and dynamic map based on position traces?”
  • 30. 30 / 32Roelof P. van den Berg Conclusion • ROMA is a proof of concept • Dynamic vector-map generation in an online fashion • Framework for map-generation evaluation
  • 31. 31 / 32Roelof P. van den Berg Conclusion – Future work • Topological and geographical map dynamics • Distributed application of ROMA Alice knows: Bob knows:
  • 32. 32 / 32 All roads lead to ROMA Friday, 6 March 2014 Roelof P. van den Berg
  • 33. 33 / 32Roelof P. van den Berg Appendix – Road Introduction
  • 34. 34 / 32Roelof P. van den Berg Appendix – Road Removal

Hinweis der Redaktion

  1. Thanks to committee: Catholijn Jonker, prof at Interactive Intelligence Leon Rothkrantz, prof at Interactive Intelligence Caroline Wehrmann, assistant prof at Science Communication Freek van Polen, researcher at Sense Observation Systems. Mentor Research performed during my internship at Sense, Daughter of Almende Design and Evaluation of a Robust On-line Map-generation Algorithm (ROMA) based on position traces Road-maps in a changing environment, such as in the next slide
  2. Translated setting -> Research Question Relevance of answering this questions Methodology to answer the question The design of the algorithm is followed by the evaluation thereof in the experiments We conclude this presentation with the discussion of results and conclusion.
  3. Accurate: Best possible representation of the world Dynamic: Quickly responds to changes in the world
  4. Young field: The main methods were described in this millenium 2003, 2006, 2009 Current literature describes how to create a map, not how to maintain Description of map dynamics and how roads decay, advances the field Map comparison for quantitative evaluation, which is rare in literature
  5. Controlled environment: simulator Determine spatial accuracy as similarity map and world Determine temporal accuracy as speed of adaption to changes Changes: Click -> introduction of roads to the world -> map generation Click -> removal of roads from the world -> road decay
  6. Road Block Simulator developed by PhD-student at Almende Click -> Provides a controllable ground-truth. Simulated world: compare ground truth world to developed world representation. Validate algorithm performance We developed both the cartography and analysis modules
  7. For validation we used Zoeterwoude-Dorp, near Leiden, as our world Navigable map representations is stored as a vector-map, consisting of positions (or nodes) with connections (or edges) between them. The vector-map is a specialized form of a graph, which is a mathematical structure commonly used to represent all kinds of problems. Vector-map has nodes representing geographical locations. Shortest path between two random points on the world. Roadblocks can be added to the world -> changes to traffic routes.
  8. Cartography module contains ROMA Robust -> responds to changes in the world Online -> provides best guess based on available measurements Map-generation Algorithm ROMA provides a vector-map representation of the world
  9. We use a sequential approach common in map generation Map dynamics is added as last step to describe how to handle changes over time ROMA is fed a stream of GPS measurements (CLICK), consisting of position information and estimate of measurement error (CLICK) The measurement error of GPS generally folows a Gaussian distribution on the horizontal plane CLICK -> It tries to match and merge these measurements to the existing map Map consists of nodes (CLICK) and edges (CLICK)
  10. Stream of measurements first goes through pre-processing -> remove outliers CLICK -> Accuracy < threshold are removed CLICK -> Inter-measurement > threshold is split Remaining measurement sequence is further processed per 6 measurements in this example, to enable look-ahead, shown in next slide
  11. Fitness-function: distance and directionality Tree search: start at E0 and further
  12. Best path sought CLICK Look-ahead No path possible CLICK
  13. V: TODO: dynamics, decay
  14. CLICK -> - Welch’s T-test - represent the same location Road decay: EXAMPLE AVG and SD of interval CLICK -> results in the inverse cumulative distribution function of a T-distribution. T-distribution was chosen because the number of measurements differs and influences the certainty of interval estimation. We also added a static initial time to live so that roads with only one measurement have a time to live estimate.
  15. Wij hebben gekozen om de twee kaarten direct met elkaar te vergelijken (topology, geometry) Kaart Zoeterwoude toevoegen
  16. CLICK -> For comparison of the developed map to the simulated world we first lay the map on the world CLICK -> We then match nodes from map to the nearest node in the world and vice versa CLICK -> Conflicts are solved by first accepting the nearest matches
  17. With the matched nodes, we can determine the true positives (CLICK) These are found when an edge is present between matched nodes in world and map CLICK -> False postives are those edges only present in the map With these two classifications, we can determine the precision (CLICK) as it is called in information retrieval Normally items from both domains are equal of size In our case, edges can differ in shape and length We therefore added scoring (CLICK) based on similarity In the world domain (CLICK), we can find the false negatives (CLICK) With false negatives and true positives we then determine recall A one value measure of spatial accuracy is obtained through the harmonic mean or F-score (CLICK)
  18. In experiment 1 we introduced new roads to the world. This was simulated by starting traffic movement. For lower amounts of traffic the map showed a rather stabilizing map quality. For higher amounts of traffic the map showed overfitting similar to overfitting in machine learning where noise gets the overhand. Overfitting is mainly found in the precision value since Recall generally stabilizes for all traffic densities. Stabilization of recall occurs within 10 minutes of all simulations. overfitting occurs earlier for larger amounts of traffic.
  19. Recall > precision in all simulations Max. Recall generally > 0.8 Max. Precision > 0.5, seems lower for more traffic CLICK -> Here you see precision decreasing after 5.5 minutes.
  20. For larger amounts of traffic we simulated road decay by adding roadblocks to the world at the moment of overfitting in experiment 1. This kind of change is visible in precision because the size of the world changes instantly. We expected the map to adapt to this change and recover to a stable value. In all experiments we observed recovery within 30 minutes and this also occured faster for larger amounts of traffic. In the simulations with large amounts of noise we measured the time to recover against a baseline without roadblocks as shown on the next slide.
  21. In the simulations with large amounts of noise we measured the time to recover against a baseline without roadblocks. Here you see the baseline experiment as a blue line, with lines for different percentages of roadblocks. CLICK -> Roadblocks are added after 5.5 minutes, the moment of overfitting. CLICK -> You see that the precision values return to baseline values within 25 minutes of road removal. What this looks like for the map is shown on the next slide
  22. After road removal, we see that the southwestern (CLICK) area loses many roads. This also happens on other locations on the map given that roadblocks are placed randomly on the world. Another observed difference is seen in the noise, or web-like structures. Because traffic takes other routes, other roads are affected by the growth of web-like structures.
  23. Practice and science Relevance for both Methodology for answers
  24. Many contributing factors: Used simulation and traffic model, applied filters, static time to live
  25. Practice and science Relevance for both Methodology for answers
  26. Potential for
  27. With this vision of the future I would like to conclude this presentation. Thank you for your time.
  28. TODO: wijs mensen op diktes van wegen -> meer verkeer
  29. SHORT IO: data belongs to user CAD: backdoor, helpdesk CL: No excessive collections DQ: errors over time (age v.s. dob)