![Page 1: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/1.jpg)
Institut für Mess- und RegelungstechnikProf. Dr.-Ing. C. StillerKarlsruhe, Germany
Description Logic for Vision-Based Intersection Understanding
Britta Hummel, Werner Thiemann, Irina Lulcheva
![Page 2: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/2.jpg)
Institut für Mess- und RegelungstechnikProf. Dr.-Ing. C. StillerKarlsruhe, Germany
• Motivation: Road recognition for toy worlds only?• DL Road Network Knowledge Base Development
– DL Tutorial– Hypothesis Space– Sensor Input
• DL Inference for Semantic Road Recognition– Deduction– Model Construction
![Page 3: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/3.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 3
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Motivation
Road RecognitionStatus Quo
• Since mid 80ies• Solved for highly restricted domains (highways)• Few work for more complex domains, then only
consideration of special cases• Geometry only, no semantics
Toy worlds?
![Page 4: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/4.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 4
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Motivation
Cannot be generalized to more complex domains
Common Approach to Road Recognition
Low-dimensionalgeometry model (clothoid, …)
2. Compare withimage cues
1. Project
3. Update
![Page 5: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/5.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 5
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Motivation
1. High-dimensional hypothesis space
2. Few features- Narrow field of view- Massive occlusions- Omitted features
3. Presence of noise- Unmodelled objects- Decreased feature quality
Problem is ill-posed!
![Page 6: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/6.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 6
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Motivation
Paradigm shift: Intersection recognition as… scene understanding task… mid high level vision task
What is needed?
1. Geometrical Model of arbitrary roads and intersections2. Massive reduction of hypothesis space size
• using prior knowledge• using large set of complementary sensor data: video object
detectors, digital map, positioning sensors, …3. Conceptual model of arbitrary roads and intersections
• explicit representation (due to intensive HMI within a DAS)4. Sound inference & retrieval services on the KB
![Page 7: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/7.jpg)
Institut für Mess- und RegelungstechnikProf. Dr.-Ing. C. StillerKarlsruhe, Germany
• Motivation: Road recognition for toy worlds only?• DL Road Network Knowledge Base Development
– DL Tutorial– Hypothesis Space– Sensor Input
• DL Inference for Semantic Road Recognition– Deduction– Model Construction
![Page 8: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/8.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 8
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Description Logic
• Decidable subset of 1st order logic• Syntax:
• Semantics: Set-theoretic
Man ∩∃hasChild.Т
to build complex expressionsC, D C ∩ D | C ∪ D |
¬C | ∃R.C | ∀R.C | T | ⊥ …
R, S R | R- | R ◦ S
ConstructorshasChildbinary relations on individualsRolesHuman(≈classes) sets of individuals ConceptsJohnobjects of the domainIndividualsExampleDescriptionName
![Page 9: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/9.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 9
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Description Logic
• Axioms form sentences
• A DL Knowledge Base consists of• Tbox: Set of terminological axioms
state general domain knowledge here• Abox: Set of assertional axioms
state knowledge about particular situation here• ( Rulebox )
John: Human (John, Emily) : hasChild
i : C(i,j) : R
Assertional Axioms
Father ≡ Man ∩ ∃hasChild.ТFather ⊆ PersonhasChild ⊆ hasDescendent
C ≡ DC ⊆ DR ⊆ S
Terminological Axioms
![Page 10: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/10.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 10
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Description Logic
• Classical DL inference services:
Retrieve all individuals which are instance of: ∃≥3hasChild ∩ ¬Female
Retrieval
John: ∃hasChild.FemaleEntailment
John: Father Classificationof Tbox and Abox
(Mother ∩ Male) is inconsistentSatisfiability Check for TBox and Abox
• Non-classical inference ….
![Page 11: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/11.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 11
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
GP1 GP2 GP3
A Geometrical Modelling
B Conceptual Modelling1. Qualitative spatial relations2. Road Network taxonomy3. Geometric Constraints 4. Road building / Semantic
Constraints
Road Network Hypothesis Space
![Page 12: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/12.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 12
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
A Geometrical Modelling
B Conceptual Modelling1. Qualitative spatial relations2. Road Network taxonomy3. Geometric Constraints 4. Road building / Semantic
Constraints
Road Network Hypothesis Space
![Page 13: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/13.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 13
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
A Geometrical Modelling
B Conceptual Modelling1. Qualitative spatial relations2. Road Network taxonomy3. Geometric Constraints 4. Road building / Semantic
Constraints
Road Network Hypothesis Space
a) Degree of overlap (RCC-Calculus)
b) Rel. orientation c) Rel. position
![Page 14: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/14.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 14
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
A Geometrical Modelling
B Conceptual Modelling1. Qualitative spatial relations2. Road Network taxonomy3. Geometric Constraints 4. Road building / Semantic
Constraints
Road Network Hypothesis Space
![Page 15: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/15.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 15
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
A Geometrical Modelling
B Conceptual Modelling1. Qualitative spatial relations2. Road Network taxonomy3. Geometric Constraints 4. Road building / Semantic
Constraints
Road Network Hypothesis Space
![Page 16: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/16.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 16
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
Arbitrary Sample
A Geometrical Modelling
B Conceptual Modelling1. Qualitative spatial relations2. Road Network taxonomy3. Geometric Constraints 4. Road building / Semantic
Constraints
Road Network Hypothesis Space
![Page 17: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/17.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 17
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
A Geometrical Modelling
B Conceptual Modelling1. Qualitative spatial relations2. Road Network taxonomy3. Geometric Constraints 4. Road building / Semantic
Constraints
Road Network Hypothesis Space
![Page 18: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/18.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 18
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
Arbitrary Road Sample
A Geometrical Modelling
B Conceptual Modelling1. Qualitative spatial relations2. Road Network taxonomy3. Geometric Constraints 4. Road building / Semantic
Constraints
Road Network Hypothesis Space
![Page 19: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/19.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 19
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
A Geometrical Modelling
B Conceptual Modelling1. Qualitative spatial relations2. Road Network taxonomy3. Geometric Constraints 4. Road building / Semantic
Constraints
Road Network Hypothesis Space
![Page 20: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/20.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 20
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
Arbitrary Sample
A Geometrical Modelling
B Conceptual Modelling1. Qualitative spatial relations2. Road Network taxonomy3. Geometric Constraints 4. Road building / Semantic
Constraints
Road Network Hypothesis Space
![Page 21: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/21.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 21
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
Sensor Data Input
1. Digital Map
2. GPS & Map Matching
3. Video/Laser-based object detectors
![Page 22: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/22.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 22
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
Symbol Grounding
[…] // r4: not(OneWayRoad)[…] // r4: has 4 Lanes […] // r4: leads to junction with 4 branches[…] // Closed World Assumption
Sensor Data Input
1. Digital Map
2. GPS & Map Matching
3. Video/Laser-based object detectors
![Page 23: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/23.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 23
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
Symbol Grounding
Sensor Data Input
1. Digital Map
2. GPS & Map Matching
3. Video/Laser-based object detectors
![Page 24: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/24.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 24
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
Symbol Grounding
[…] // i1: RoadCurb […] // i2: StraightAheadArrow[…] // i3: DividerMarking50-20[…] // state spatial relations […] // Open World Assumption
Sensor Data Input
1. Digital Map
2. GPS & Map Matching
3. Video/Laser-based object detectors
![Page 25: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/25.jpg)
Institut für Mess- und RegelungstechnikProf. Dr.-Ing. C. StillerKarlsruhe, Germany
• Motivation: Road recognition for toy worlds only?• DL Road Network Knowledge Base Development
– DL Tutorial– Hypothesis Space– Sensor Input
• DL Inference for Semantic Road Recognition– Deduction– Model Construction
![Page 26: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/26.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 26
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.
• not a one way road• has 4 lanes
Road Network KB
1. Draw all conclusions, given the KB (i.e. prior knowledge & sensor data)
Deduction
2. Generate all intersection hypotheses
Model construktion
Video-based hypothesis test / Deformable Model matching
Inference
![Page 27: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/27.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 27
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
Inference
Abox A1
Abox An
. . .H =
1. Draw all conclusions, given the KB (i.e. prior knowledge & sensor data)
Deduction
2. Generate all intersection hypotheses
Model construction
Video-based hypothesis test / Deformable Model matching
![Page 28: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/28.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 28
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Road Network KB
Inference
Symbol Grounding
Abox Ak1. Draw all conclusions, given
the KB (i.e. prior knowledge & sensor data)
Deduction
2. Generate all intersection hypotheses
Model construktion
Video-based hypothesis test / Deformable Model matching
![Page 29: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/29.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 29
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Demo: Video
geometry model generated from 1st order knowledge base containing ground truth data
![Page 30: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/30.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 30
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Demo: Video
![Page 31: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/31.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 31
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Summary
Motivation:• Road recognition works in toy worlds only• A massive augmentation of prior knowledge is needed
Knowledge Representation:• Introduction of DL as a formal, explicit, and readable knowledge
representation formalism for real-world high-level scene interpretation• Development of a DL RoadNetwork KB
– High-level conceptual constraints – Map, Positioning, Video/Laser sensors
Inference:• Deductive reasoning for querying the KB• Model Construction for returning the set of intersection hypotheses• Natural handling of partial observability, differing abstraction layers and
incremental additions
} constrain hypothesis space
![Page 32: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/32.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 32
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Outlook
• Hypothesize & Test• Migration to probabilistic logic (MLN, SLP)• Rule Learning from Training Data
![Page 33: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/33.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 33
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Thanks…
Thanks for your attention.
![Page 34: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/34.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 34
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Example: Proof of Result 5
![Page 35: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/35.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 35
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Example: Proof of Result 5
![Page 36: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/36.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 36
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Example: Proof of Result 5
![Page 37: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/37.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 37
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Example: Proof of Result 5
![Page 38: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/38.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 38
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.
// A lane with a straight ahead arrow is a straight ahead lane[…]// Right neighbors of straight ahead lanes are straight ahead or right turn
lanes only.[…]// Bicycle lanes do not occur between lanes of the same turning lane type[…]
Example: Proof of Result 5
![Page 39: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/39.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 39
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Example: Proof of Result 5
![Page 40: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/40.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 40
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.
• Intermediate result:
As r1 is a OneWayRoad towards the junction, r4 cannot have a right turn lane.
Example: Proof of Result 5
![Page 41: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/41.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 41
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.
==>
• Closed World Assumption for map data
Example: Proof of Result 5
![Page 42: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/42.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 42
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Hypothesize&Test
Evidence (ABOX):RightTurnArrow(arrow1);
SingleLongDashedDivider(divi-der1);
Hypothesization:Lane(lane1);hasMarking(
lane1, arrow1);
Hypothesization:isConnectedWith-
RightDivider(divider1);
isConnectedWith-LeftDivider(divider2);
Hypothesization:Road(road1);isPartOf(lane1,
road1);
Hypothesization:Lane(lane2);
hasRightNeighbor(lane1,lane2);
Deduction:isPartOf(lane2,
road1);RightTurnLane(
lane2);
Deduction:MarkedRoad(road1);
DividerMarking(
divider2);
PhysicalDivider(
divider2);
non-trivialHypothesization:8a.
8b.
1. 2. 3. 4.
5. 6. 7.
![Page 43: Description Logic for Vision-Based Intersection Understanding](https://reader031.vdocuments.site/reader031/viewer/2022012301/61e1db81b5949021010543b9/html5/thumbnails/43.jpg)
Universität Karlsruhe (TH), GermanyInstitut für Mess- und Regelungstechnik
Prof. Dr.-Ing. C. Stiller
Britta Hummel 43
©2007 Alle R
echte einschließlich Patentier-, Kopier-und W
eitergaberechte bei uns.Hypothesize&Test
...... ......
model complete
8a.
8a.n.