An ontology representation of indoor spaces useful for electronic navigation tools for the vision impaired community is presented here. The specific needs and how the building environment, features and their attributes contribute to safe and efficient navigation is explored from an orientation and mobility perspective, comparing with standard building models. The limitations of standard building models in addressing the unique requirements of the vision impaired community is addressed in this approach. The ontology forms the basis for defining a navigation model suitable for the vision impaired community, serving the needs of different groups of vision impaired individuals to derive user centric, context aware route determination and navigation instructor formation.
Navigating in large complex buildings becomes a necessity in fulfilling life needs in
the modern civilization for everybody, including vision impaired (VI) individuals.
However, with impaired vision, the human orientation and mobility abilities are heavily
affected and wayfinding becomes an extensively challenging task. Electronic indoor
navigation aids specific for the VI community are being developed complementing
traditional aids such as white canes or guide dogs, to support the navigation needs of the
VI community. Navigation is a process which requires constant negotiation with the
environment. Therefore the built environment representation is considered as a main
module in such travel aids [
The environmental features which can be easily negotiated with vision become serious safety concerns for VI navigators. For example, fine grained information, such as direction of a door opening, virtually irrelevant for a sighted person, becomes useful information while visually rich landmarks, notable by a sighted person, become meaningless for a VI person. Therefore, how the built environment is modelled and represented becomes an important step for realising a successful VI navigation aid.
Though common building models support general navigation needs, they have limitations related to VI navigation; the type of features modelled may not capture required information such as barriers or landmarks; the level of detail would not be enough for safety concerns; the representations may not suitable for satisfying VI mobility needs. In the absence of properly defined knowledge on building information requirements for VI navigation, this work defines a VI indoor space ontology aimed at providing a sufficient level of information for navigation models and navigation tools. 2
The concepts of building are well documented from a construction and architectural
point of view via national building codes (ex: Australian building codes [
An example of how IndoorGML can be extended for VI navigation is presented as
a use case in [
Ontologies are an effective approach to represent the knowledge of a domain using
objects, relationships and descriptions so it can be used by different resources, perform
semantic queries on the information and infer new knowledge. There are studies on
using an ontological approach in general navigation modelling [
Skills, techniques and guidelines, collectively called Orientation and Mobility (O & M)
guidelines are a key resource for a VI individual in learning independent navigation
skills. Therefore, first, the information required from the environment in VI navigation
activity is identified from the O & M point of view. For that the O & M domain is
explored, referring to standard O & M guidelines [
The listing below summarizes the main environmental information expected from a VI navigation tool by a typical VI user as per the O & M domain. 1. Identification of obstacle at front from ground level to head height and over a wide enough area horizontally to cover the width of the traveler's body 2. Identification of the nature of the walking path and the surface nature (e.g. slippery, tiled, carpeted etc.) including abrupt elevation changes with safety concerns(e.g. split floors, steps, down stairs ) 3. Detect objects at sides, including doorways, walls etc. which forms shoreline on either side of the path 4. Detect walls at side of the walking areas which can be used for wall training 5. Information to know about heading and direction of movement, distant landmarks to support maintaining the course 6. Information to allow the traveller to build up a mental map, image, or schema for the chosen route to be followed, including turns and other discontinuities 7. Landmarks in immediate surrounding referring to specific building features 8. Identification of challenging transit features between floors (e.g. staircases) 9. Identification of connecting features within floors such as doors with sufficient details for safely opening
Items 1-3, 5-7 are based on [
Based on the requirements defined above, the building features needed to be modelled are identified and presented in the Table 1. Contrast and light sensitivity also can be reduced with certain forms of vision loss and affects environment sensing hence included. Most of the features identified have architectural meaning but need some additional information explicitly.
Requirement in VI terms Possibility to use shorelining techniques
Wall-trailing techniques Know about 3 fixed or temporary hazards on the way abrupt elevation changes Floor type Escalators elevators staircases ramps 10 large open spaces 11 human traffic 12 doors and openings 13 food serving places served 14 landmarks a.visual landmarks b.tactile landmarks c.olfactory landmarks d.audio landmarks
Contrast levels of a the area Lighting levels of a the area
User preference would vary; high contrast setting would enable functional vision ; low contrast areas may affect safety
User preference would vary; better lighting enable functioning vision
The features identified above are conceptualized based on the idea of VI navigation in the indoor environment. For example, a landmark should be alone the path even to be detected by a tactile mean; a wall should be presented along the navigation path to it to be used for shorelining or wall-trailing.
The notion of cellular space, which is adopted by IndoorGML [
Accordingly, we define two hypothetical disjoint constructs, VICellSpaceSection and VICellSectionBoundary, which can be linked to actual cell spaces of the indoor environment via a grid based partitioning for the indoor space. VICellSectionBoundary, which constitutes of both navigable and non-navigable boundaries define the boundary of the cell section. 5
VIindoorLandmark: any indoor feature which can support developing mental map or establish orientation for a VI navigator VIIndoorObstacle: any indoor feature which can become a safety concern for a VI navigator VIIndoorFloorConnectionSpace: any indoor construct used to connect two levels of floors or split floors VIIndoorOpening: an opening in a wall or boundary of a passage which may or may not be having a door VIIndoorFloorSection: floor of an indoor walking area VIIndoorWallSection: wall bounding any indoor space VIIndoorBuildingInstallation: any fixture such as fire equipment in indoor which are relatively fixed VIIndoorfurniture: furniture items in indoor space.
Some supportive classes are: VIIndoorMaterial: define the material or textures use in indoor floor sections or walls sections VIContrastValuePartition: define the contrast levels which can affect the environment sensing with four variations.
As staircases, elevators, ramps, escalators and steps are a major challenge for VI community and need more details each of them are modelled as separate classes as subclasses of VIFloorConnectingSpace. How these features and attributes are represented in compared standards are considered to reuse of the vocabulary where possible. Three obstacle types are defined based on the part of the body it would affect and from which side it can affect. VIObstacle can be either a VIBuildingInstallations, VIBuildingFurniture, VIHumanClutter or a VIRandomItems. 6
The ontology defined can be used in different ways to support the concerned
community. First, formal queries based on SPARQL or GeoSPARQL[
Third, VI navigation models could be derived based on the ontology as the foundation. As the building space is conceptualized with a space model coupled with se-mantic information and relationships, a VI navigation focused building model can be defined which can later be utilised for navigation modelling. Such a model would be useful in route determination and navigation instruction formation which would include landmarks, obstacles and other VI user specific constructs as inputs.
Though the ontology is attempted to be comprehensive, there are features which are
not addressed; some of the features modelled would have attributes which vary with
time (e.g: human clutter); the requirements of the VI community as well as building
features can vary between geographic areas (e.g: tactile markers).The proposed
ontology is influenced by the IndoorGML specification [
Finally, comprehensive rules should be derived and tested to verify the ontology and check the consistency before it is used for actual instances of complex building data. The next stage of this work is to look at (a) formal testing and deriving route determination rules with the defined ontology (b) deriving navigation model for VI people, especially as an extension for IndoorGML using the derived ontology as the basis. 7
A building ontology specifically serving VI navigation needs is defined in this study. The unique requirements of the VI community is identified by analysing the VI information requirement from O & M perspective and conceptualized the indoor areas accordingly. It is capable of representing building feature semantics and use hypothetical constructs to model navigable areas for VI users. The ontology defined would be useful in different ways to develop more useful indoor navigation tools for VI community and open for enhancements as well.
This study is initially supported by COPRS scholarship from the Curtin University, as per its affiliation with SLIIT, Sri Lanka and now being supported by the Research Training Program (RTP), Australian Government. Support provided by the Association for the Blind,WA and Ms. Manique Gunaratne , EFC’s ICT Training Centre for the Visually Impaired, Colombo are kindly acknowledged.