The Architecture of Future Automotive Applications based on Web - - PowerPoint PPT Presentation

▶

Aug 23, 2022 481 likes •712 views

The Architecture of Future Automotive Applications based on Web Technologies Luka Bradesko, Marko Grobelnik, Andreas Harth, Achim Rettinger and Sebastian Speiser (JSI, AIFB, KSRI) W3C Workshop on Web and Automotive Rome, Italy November 13 th

SLIDE 1

Karlsruhe Service Research Institute (KSRI), Institute of Applied Informatics and Formal Description Methods (AIFB)

www.kit.edu

KIT – University of the State of Baden-Württemberg and National Research Center of the Helmholtz Association KIT – University of the State of Baden-Württemberg and

The Architecture of Future Automotive Applications based on Web Technologies

Luka Bradesko, Marko Grobelnik, Andreas Harth, Achim Rettinger and Sebastian Speiser (JSI, AIFB, KSRI)

W3C Workshop on Web and Automotive Rome, Italy November 13th 2012

SLIDE 2

Agenda § Motivation § Overview § Memory: Virtual Knowledge Base § Interoperation Layer § Conclusions

15.11.2012 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012 2

SLIDE 3

MOTIVATION

The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012 15.11.2012 3

SLIDE 4

Motivation § Besides reaching B from A, a driver has other goals, e.g.,

§ Adjusting delivery routes according to traffic § Finding cheapest petrol station along the route § Handling the radio and phone

§ Result:

Source: http://www.adslogistics.com/blog/bid/53021/Avoid-the-Three-Types-of-Distracted-Driving

15.11.2012 4 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

SLIDE 5

Motivation A rare example of a driver solely concentrated on driving …. but …

Source: http://www.f1-site.com/wallpapers/2009/f1/abu_dhabi/drivers/photo-f1-team-driver-wallpaper-2009-3.jpg

15.11.2012 5 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

SLIDE 6

Motivation A lot of assistants help out in the background with a convenient speech interface

Source: http://www.f1-site.com/wallpapers/2009/f1/abu_dhabi/drivers/photo-f1-team-driver-wallpaper-2009-3.jpg

15.11.2012 6 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

SLIDE 7

Motivation § Formula 1 solution is impractical for widespread use: costs of about 1,000 USD / kilometer

source: http://www.faz.net/aktuell/sport/formel-1/was-kostet-die-formel-1-ein-teures-rennvergnuegen-1257804.html

§ Our proposed solution: Driver assistant systems based on

§ Speech recognition § Virtual knowledge base integrating data from the web § Logical inferences § Statistical learning

15.11.2012 7 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

SLIDE 8

OVERVIEW

The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012 15.11.2012 8

SLIDE 9

High-level Architecture

Logic Reasoner Speech I/O Stream I/O Cyc Data I/O LOD Memory Adaptive Learner Interoperation Layer

15.11.2012 9 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

SLIDE 10

Overview of Architecture § Speech I/O: Conversational interface

§ use patterns only to help us converting text to logic § reasoning engine to control the conversation § replies: use Cyc’s reasoner logic-to-language translation § reasoning lets us get rid of the need for lots of prepared patterns

§ Stream I/O

§ constant stream of data and events § source: car sensors, web streams, user sensors § representing information about car, environment, user

15.11.2012 10 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

SLIDE 11

Data I/O: Linking Open Data Cloud § Linked Data (RDF data accessible via HTTP lookups), 2006 § Yearly growth rates of ~200% § Many datasets, covering descriptions of millions of entities § Large number of interlinked distributed disparate small data sources rather than single-source single-organisation knowledge bases

15.11.2012 11 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

source: http://lod-cloud.net/

SLIDE 12

Combining and Using I/O Channels § All the data is put in a virtual knowledge base (more in a minute) § Interoperation layer relates data from different channels (more in two minutes) § Based on the data infer new knowledge

§ Logic reasoner (exact methods) § Learning component (statistical and heuristic methods)

15.11.2012 12 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

SLIDE 13

MEMORY: VIRTUAL KNOWLEDGE BASE INTEROPERATION LAYER

The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012 15.11.2012 13

SLIDE 14

Memory: Virtual Knowledge Base § Based on Linked Data and REST principles § Basic abstraction: a resource

§ different representations, e.g., a POI as 3D XML, RDF, or JPEG § resources have references (links) to other related resources, e.g., leading to the next step in a series of operations § support standard operations: CRUD

§ A resource has an identifier (HTTP URI) § The identifier specifies a way to access information about the resource (performing a HTTP lookup) § The information is in standardised format (an RDF graph describing the resource using its identifier)

15.11.2012 14 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

SLIDE 15

Interlinking in Linked Data § Establishing equivalencies across sources, e.g.,

car:currentDriver owl:sameAs facebook:JohnDoe  (car: stands e.g. for http://localhost/...)

§ All statements about JohnDoe also apply to current driver and vice versa § Using URIs across services, e.g.,

car:car :location car:point  car:point foaf:based_near wikipedia:Rome

15.11.2012 15 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

SLIDE 16

Streams, Services and Compositions § Streams, either (depending on frequency of updates)

§ just do not close HTTP connection and continue to list § pull “stream source” regularly

§ Services / dynamic data / data with limited access patterns

§ integrate with Linked Data

§ How to build

§ applications, § compositions, § workflows

§ based on REST resources?

15.11.2012 16 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

SLIDE 17

15.11.2012 17 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

Linked Services § Linked Services:

§ service approach based on web architecture (REST) § describing web resources with RDF/Notation3 aspects: input, output, relation between input and output) § linking between: services, descriptions, data in input and output

§ CRUD operations on resources

Barry Norton and Reto Krummenacher. Consuming Dynamic Linked Data. In Proceedings of the 1st International Workshop on Consuming Linked Data (COLD’10), 2010. Sebastian Speiser and Andreas Harth. Integrating Linked Data and Services with Linked Data Services. In Proceedings of the 8th Extended Semantic Web Conference (ESWC’11), 2011.

SLIDE 18

15.11.2012 18 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012

Production Rules for Linked Services § Decentralised linking between resources § Use production rules to specify composition of resources § Depending on current state of knowledge base (KB):

§ invoke new services (add their output to KB) § find links to new services in the KB

Steffen Stadtmüller, Andreas Harth. „Towards Data-driven Programming for RESTful Linked Data“. Workshop PSW, ISWC 2012. Example: navigate to gas station

car location new location list of gas stations programmed GPS system

locate navigate cancel lookup

SLIDE 19

CONCLUSIONS

The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012 15.11.2012 19

SLIDE 20

Conclusions § Using web technologies to provide assistant system to driver’s at reasonable cost § Speech interface based on logic reasoning § Web technologies enable interoperability

§ new sources can be added on demand § pay-as-you go for integration § we can tap the long tail of sensors and data sources

15.11.2012 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012 20

SLIDE 21

Thank you for your attention. www.ksri.kit.edu www.aifb.kit.edu Supported by the iZEUS project (German Federal Ministry of Economics and Technology)

15.11.2012 The Architecture of Future Automotive Applications based on Web Technologies - Sebastian Speiser - W3C Web and Automotive 2012 21

SLIDE 22

Karlsruhe Service Research Institute (KSRI), Institute of Applied Informatics and Formal Description Methods (AIFB)

www.kit.edu

KIT – University of the State of Baden-Württemberg and National Research Center of the Helmholtz Association KIT – University of the State of Baden-Württemberg and

www.kit.edu

The Architecture of Future Automotive Applications based on Web Technologies

Luka Bradesko, Marko Grobelnik, Andreas Harth, Achim Rettinger and Sebastian Speiser (JSI, AIFB, KSRI)

W3C Workshop on Web and Automotive Rome, Italy November 13th 2012

Agenda § Motivation § Overview § Memory: Virtual Knowledge Base § Interoperation Layer § Conclusions

MOTIVATION

Motivation § Besides reaching B from A, a driver has other goals, e.g.,

§ Adjusting delivery routes according to traffic § Finding cheapest petrol station along the route § Handling the radio and phone

§ Result:

Motivation A rare example of a driver solely concentrated on driving …. but …

Motivation A lot of assistants help out in the background with a convenient speech interface

Motivation § Formula 1 solution is impractical for widespread use: costs of about 1,000 USD / kilometer

§ Our proposed solution: Driver assistant systems based on

§ Speech recognition § Virtual knowledge base integrating data from the web § Logical inferences § Statistical learning

OVERVIEW

High-level Architecture

Logic Reasoner Speech I/O Stream I/O Cyc Data I/O LOD Memory Adaptive Learner Interoperation Layer

Overview of Architecture § Speech I/O: Conversational interface

§ use patterns only to help us converting text to logic § reasoning engine to control the conversation § replies: use Cyc’s reasoner logic-to-language translation § reasoning lets us get rid of the need for lots of prepared patterns

§ Stream I/O

§ constant stream of data and events § source: car sensors, web streams, user sensors § representing information about car, environment, user

source: http://lod-cloud.net/

Combining and Using I/O Channels § All the data is put in a virtual knowledge base (more in a minute) § Interoperation layer relates data from different channels (more in two minutes) § Based on the data infer new knowledge

§ Logic reasoner (exact methods) § Learning component (statistical and heuristic methods)

MEMORY: VIRTUAL KNOWLEDGE BASE INTEROPERATION LAYER

Memory: Virtual Knowledge Base § Based on Linked Data and REST principles § Basic abstraction: a resource

§ different representations, e.g., a POI as 3D XML, RDF, or JPEG § resources have references (links) to other related resources, e.g., leading to the next step in a series of operations § support standard operations: CRUD

§ A resource has an identifier (HTTP URI) § The identifier specifies a way to access information about the resource (performing a HTTP lookup) § The information is in standardised format (an RDF graph describing the resource using its identifier)

Interlinking in Linked Data § Establishing equivalencies across sources, e.g.,

car:currentDriver owl:sameAs facebook:JohnDoe (car: stands e.g. for http://localhost/...)

§ All statements about JohnDoe also apply to current driver and vice versa § Using URIs across services, e.g.,

car:car :location car:point car:point foaf:based_near wikipedia:Rome

Streams, Services and Compositions § Streams, either (depending on frequency of updates)

§ just do not close HTTP connection and continue to list § pull “stream source” regularly

§ Services / dynamic data / data with limited access patterns

§ integrate with Linked Data

§ How to build

§ applications, § compositions, § workflows

§ based on REST resources?

Linked Services § Linked Services:

§ service approach based on web architecture (REST) § describing web resources with RDF/Notation3 aspects: input, output, relation between input and output) § linking between: services, descriptions, data in input and output

§ CRUD operations on resources

Production Rules for Linked Services § Decentralised linking between resources § Use production rules to specify composition of resources § Depending on current state of knowledge base (KB):

§ invoke new services (add their output to KB) § find links to new services in the KB

Steffen Stadtmüller, Andreas Harth. „Towards Data-driven Programming for RESTful Linked Data“. Workshop PSW, ISWC 2012. Example: navigate to gas station

car location new location list of gas stations programmed GPS system

locate navigate cancel lookup

CONCLUSIONS

Conclusions § Using web technologies to provide assistant system to driver’s at reasonable cost § Speech interface based on logic reasoning § Web technologies enable interoperability

§ new sources can be added on demand § pay-as-you go for integration § we can tap the long tail of sensors and data sources

Thank you for your attention. www.ksri.kit.edu www.aifb.kit.edu Supported by the iZEUS project (German Federal Ministry of Economics and Technology)

www.kit.edu

The Architecture of Future Automotive Applications based on Web Technologies

Luka Bradesko, Marko Grobelnik, Andreas Harth, Achim Rettinger and Sebastian Speiser (JSI, AIFB, KSRI)

W3C Workshop on Web and Automotive Rome, Italy November 13th 2012

car:currentDriver owl:sameAs facebook:JohnDoe  (car: stands e.g. for http://localhost/...)

car:car :location car:point  car:point foaf:based_near wikipedia:Rome