A Humancentered Semantic Navigation System for Indoor Environments

1
A Human-centered Semantic Navigation System for
Indoor Environments
V. Tsetsos, C. Anagnostopoulos, P. Kikiras, T.
Hasiotis and S. Hadjiefthymiades

• Pervasive Computing Research Group,
• Communication Networks Laboratory (CNL),
• Dept. of Informatics Telecommunications,
• University of Athens

International Conference on Pervasive Services
(ICPS05) July 11-14 2005 Santorini, Greece
2
Presentation Structure

• Introduction
• System Design
• Conclusions

3
Pedestrian Navigation Systems

• Navigation a typical Location Based Service
• Indoor vs. Outdoor
• Outdoor Navigation has attracted more interest,
  partly due to GPS proliferation
• Indoor Navigation research is driven by robot
  navigation
• Pedestrian Indoor Navigation
• Museum guides
• Special solutions for people with disabilities
• Semantic LBS A new human-centered approach
• Intelligent service provision based on
  ontological knowledge representation and hybrid
  location modeling
• Human-centered services, suitable for people with
  disabilities
• Can be deployed in smart environments

4
Motivation

• Complex and unknown built environments cannot be
  easily explored
• People with disabilities face additional
  difficulties and put extra effort when following
  paths that eventually become non-traversable
• Built environments are associated with rich
  semantics which may lead to intelligent services
  if appropriately exploited
• OntoNavs goal to assist the path selection and
  E2E user guidance by exploiting the semantics of
  indoor spaces

5
Use Case (I)
Destination
5 walkable paths
6
Use Case (II)
Destination
2 traversable paths
Best Traversable Path is not the shortest
walkable path
7
Presentation Structure

• Introduction
• System Design
• Conclusions

8
OntoNav Architecture

• Navigation Service (NAV)
• Geometric Path Computation Service (GEO)
• Semantic Path Selection Service (SEM)

9
System Functionality (I)
User-compatible graph
Building graph
Building blueprints (GIS)
Data Migration
Spatial DB
INO instances
User-compatible INO instances
Indoor Navigation Ontology (INO)
User profile (capabilities)
10
System Functionality (II)
User-compatible graph
User profile
GEO
SEM
NAV
k shortest paths
Best Traversable Path
User-compatible INO instances
User and destination locations
11
Indoor Navigation Ontology (INO) I

• Represents complex built environments from a
  navigation perspective
• Imports concepts from indoor location ontology
• Building, Floor, Room, Corridor,
• no well-established indoor location ontology
  exists ? merging and extensions of existing
  indoor location models are pursued
• INO is currently in evaluation phase

12
Indoor Navigation Ontology (INO) II
13
User profiles

• Describe the capabilities and preferences of
  users
• A user profile (UP) contains
• Physical navigation rules (e.g., wheelchair)
• Perceptual navigation rules (e.g., child,
  illiterate)
• Routing preferences (e.g., calendar-based)
• A user typically selects a predefined UP and
  further adjusts it
• UPs are implemented as sets of rules that refer
  to INO vocabulary and are applied to INO
  instances
• e.g., if user x cannot walk and path p contains a
  vertical passage v of type stairs then p is
  excluded

14
NAV Service

• Provides the interface between end-users and
  OntoNav
• Receives user requests
• Retrieves user position and location of
  destination
• Handles path presentation issues

15
GEO Service

• Inputs
• a planar graph that accumulates the floor
  sub-graphs that are traversable by a specific
  user
• user and destination locations
• Output
• k shortest paths
• Edgescorridor segments, verticesexits and
  passages (i.e., each location is reduced to a
  set of exits or passages)
• Performs hierarchical clustering in the graph on
  a floor basis for more efficient path discovery
• We cannot promote only the shortest path as it
  may not satisfy all user preferences

16
SEM Service

• Inputs
• user profile
• k shortest paths
• Output
• best traversable path (BTP) and its anchors
• BTP is the path that
• satisfies the routing preferences of the user
  profile
• contains adequate number of appropriate anchors
  (landmarks) for its presentation
• An anchor is appropriate if it satisfies the
  physical and perceptual navigation rules of the
  user
• Example for a blind or illiterate user the BTP
  should contain voice-enabled anchors

17
Implementation Issues

• OntoNav is currently being developed using
• Web Ontology Language (OWL-DL) - navigation
  ontology
• Semantic Web Rules Language (SWRL) - user
  profiles
• SweetRules v2.1 - SWRL rules engine
• Racer, Pellet OWL reasoning engines
• PostGIS spatial database
• JUNG Java graph library

18
Presentation Structure

• Introduction
• System Design
• Conclusions

19
Added Value of OntoNav

• A purely user-centric navigation system, that
  adheres to the Inclusive Design paradigm
• Hybrid location model (geographic and symbolic)
  that
• Enables more advanced interpretations of distance
  than the Euclidean one
• Introduces user-defined quality metrics to the
  path selection process
• Suitable for intelligent context-aware
  environments

20
Future Work

• Inference of user status for assistance during
  the navigation process
• e.g., identification of lost, wandering,
  stationary, or deviated users
• Study the applicability of path caching
  techniques
• Deployment on our campus facilities for
  evaluation
• URL http//p-comp.di.uoa.gr/projects/ontonav