A Computer-Based Navigation System Tailored to the Needs of Blind People

Bettina Pressl, pressl@tugraz.at
(Manfred Wieser, wieser@geomatics.tu-graz.ac.at)
from Graz University of Technology, Austria


Slide 1: Content

First I would like to introduce you to the project, the intention, the concept.

Then I will describe the components of the navigation system:
These are a navigable map, the route planning algorithm, positioning, map matching and guidance.

Finally I will give an outlook on a following project.


Slide 2: The project

The project is called PONTES.
It is an abbreviation for Positioning and navigation of visually impaired pedestrians in an urban environment.

Financial support is given by the Austrian Ministry of Transportation.

We are the Institute of Navigation and Satellite Geodesy from Graz University of Technology.

For our project, we have a cooperation with the Styrian Association of the Blind and Visually Impaired. So from the beginning blind people
are involved in the project for defining the user requirements and for testing the prototype.

Another cooperation is with Vectronix AG from Switzerland. From this company we have in use a positioning unit for pedestrians.


Slide 3: The intention

The intention was to create an application with social relevance. It is an overall outdoor-navigation concept with the components information, navigation and a man-machine-interface.
The architecture of the system is tailored to the specific user group.
Some underlying issues are:
The aim of the project was to develop a demonstrator prototype and a proprietary software.
Another task was a feasibility study and the evaluation of the consumer acceptance. Both had very promising results.


Slide 4: The concept of the system

In the center there is the geodatabase which represents the navigable map containing the whole information needed.
Based on this map, route planning and map matching can be done.
The map matching module also needs the current position of the user, which is determined by the positioning module.
Map matching and route planning are the basic modules for guidance.
The man-machine interface communicates the guidance information to the user.
Now I will describe the components.


Slide 5: Navigable map 1/2

It is the digital answer to tactile maps.
The modeling of the urban environment is needed for routing and guidance, map aiding and map matching.
The digital map has to be geometrically accurate, topologically consistent, thematically correct, up-to-date, and complete.
The data has to be standardized and stored efficiently.


Slide 6: Navigable map 2/2

The navigable map is created based on a digital city map.
Based on that, we created a digital path network which is stored in a database.
The path network consists of nodes and edges (crossings and paths), polygon point (obstacles), and points of interest.


Slide 7: Routing 1/2

Based on the navigable map route planning can be done.
The mathematical representation of the path network is a valuated directed graph.
A specific cost function contains geometry and safety aspects.


Slide 8: Routing 2/2

For route planning we implemented Dijkstra's shortest path algorithm.
But not only the shortest path can be computed, much more important for blind people is the safest route.
So we included a specifically developed cost function. With this function, especially road crossings without traffic lights can be avoided.
With the output of the routing, a maneuver list is created. It contains the object type, the distance walked, a main guidance information and additional guidance information.


Slide 9: Positioning 1/2

As far as positioning is concerned, there are very high quality requirements.
The accuracy should be within the "stick range".
There should be availability anytime and everywhere and a top-level integrity.

So a sensor fusion is used. It includes satellite-based positioning and dead reckoning, which means course determination and step detection.


Slide 10: Positioning 2/2

We have in use a pedestrian navigation module from Vectronix AG.
This module contains a GPS-receiver, a digital compass unit, an accelerometer triad, and a barometric altimeter.
The advantage of the system is that it allows the bridging of GPS-data gaps with dead reckoning.


Slide 11: Map matching

The map matching module projects the determined position to the digital path network. We developed a sophisticated software which is based on an edge-to-edge matching.
This is realized by an affine transformation with least squares alignment.


Slide 12: Guidance 1/2

The guidance instructions are based on the navigable map as well as the position determination.
The navigable map is followed by route planning and the maneuver list, which defines "how" to guide the user.
The positioning is followed by map matching and route checking, which defines "when" to give the next guidance instruction to the user.


Slide 13: Guidance 2/2

The system guides the user by supplementing usual tactile aids.
The user gets obstacle warnings. E.g. warnings of traffic signs or post poxes.
There is a maximum of two hints per maneuver, otherwise there will be too much information.
The system also includes warnings of off-road and off-route situations.


Slide 14: Man-machine-interface (MMI)

The MMI is not yet tailored to the needs of blind people. We will realize this in our following project.
Possibilities and problems:
Input:
Keypad: too big for mobile applications
Buttons: Good, but addresses for the routing could not be entered directly
Speech recognition: problem to recognize speech in noisy surroundings

Output:
Earphones: Blind people do not like it, because wearing it, they are not able to hear the background noise, which is important for their orientation.
Vibrating wristbands: Wearing it on both hands the information "left" and "right" can be communicated. By varying the intensity obstacle warning can be done.


Slide 15: Outlook

The future vision is to use the system for indoor navigation and miniaturize the system.
In our following project public transport will be included. It is a challenge for the positioning-module, because in busses or trams you have neither GPS-signals nor you can do step detection.
Moreover we will realize a virtual training at home, and a module, that allows the blind user to add data by himself.


Slide 16: Discussion and Thanks