Navigation system for the Blind on UF Campus

Report due 4-28-99

Steve Moore

Suppose you are blind person who works in a campus setting, i.e. a cluster of buildings all linked by various pathways containing many obstacles. If you wanted to travel from one location on campus to another, a list of things to worry about (consider) such as the following may begin to grow in your head:

• How soon do I need to get there
• How do I get there

• Should I take a sheltered route

• Which route has the fewest obstacles

• Which route has the least amount of traffic during this time of day

• What is the weather like

• Who will help me get there

Once you get there:

When you enter a room on campus, you wonder about its layout,

where the chairs are which way are they facing, where the black board is, where the desk is?

Also you might like to know where the nearest restroom is, or nearest water fountain is.

The following is the working design of a system to address these questions, and others encountered on the way.

The system should be able to keep a blind traveler on a path from one location to another, and keep the traveler "comfortable" with the path they are traversing.

I envision the following as a possible scenario…

Computer, goto room 227, Anderson hall. The computer can now describe to me, if I so desire, the entire route that I will travel to get from my current location, to Anderson 227. Out the door, turn to your left, you will be in a hall way about 7 feet in width. You will walk down the hall way about 8 paces. You will turn to your right and walk about 9 paces, there is a waste can at about 6 paces on your right, … you should be able to stop this monologue and start it up again if you want form your current location onward towards your destination. Once you are on the path the system will keep you on the path (within a certain tolerance, different levels for different path surroundings) with verbal prompts, such as move to your left, move to your right. This correction prompt can be enhanced with headphones, using the left and right earphones separately. Also when traversing the path the system should announce the presence of upcoming obstacles and hazards, it should also be able to describe the size, shape, duration of the obstacle. We should determine what vocabulary is appropriate to describe objects to a person who has never had sight. To give a person a feel for the layout of the path they are on and where buildings are on campus, the system should announce buildings that are being passed by, and be able to give a description of their function. Here we like to have different granularities of descriptions available.

To put current distance from an object into voice, we can use different levels of volume.

Using a GPS receiver one can determine their coordinates on the earth, however with errors in the range of ± 50 meters. To get readings with errors in the centimeters we need to use Differential GPS. The DoD injects error into the signal being broadcast by its NAVSTAR satellites, also since to determine ones position the receiver essentially measures the amount of time it takes the satellite signal to propagate to its location, different atmospheric conditions can contribute to some error. To do Real Time DGPS a fixed receiver has to be set up that knows its coordinates precisely, it can then continuously monitor satellites and determine the error, and broadcast this information out to mobile receivers. So one can set up there own base station, or The Coast Guard has base stations along the coast that broadcast this information and is used at the Geoplan center on the UF campus. Also there is a receiver set up in a tower on the West Side of Gainesville that broadcasts out error correction information to subscribers, by way of a pager interface to a mobile GPS receiver. Other factors to be considered in a GPS receiver are the number of channels it can monitor simultaneously, and the number of satellites it can track. To get a reading a receiver needs the signals from 4 different satellites, if it can choose satellites that are spaced out appropriately it can reduce the error. Thus 4 or more channels and tracking of 8 satellites is desirable.

Using a compass a traveler can be told which direction they are traveling and if it is the appropriate direction. They can also use this to orient themselves in rooms.

A blind person who wants to get from one location to another,

A Differential Geographical Positioning System receiver: by monitoring the signal from at least 4 NAVSTAR satellites orbiting the earth and obtaining error correction data from a "nearby" GPS base station the receiver can determine its position on the earth with sub-meter accuracy within every 5 seconds,

An electronic compass,

A Personal Computing Assistant interfaced with, the traveler (verbally), DGPS receiver and the compass.

To construct a path from one location to another I think we would need to go out and determine an appropriate route to take, and then with the DGPS receiver mark out points along the path.

For Curves in the path I would map out a sequence of straight-line segments that fit the curve, this will make it simpler to determine a traveler’s distance from the path using perpendicular distance from a point to a straight line.

We need to make sure that we associate obstacles along the route with a point, or pair of points for beginning and end of such things as bricked planters that parallel a path segment.

We need to associate buildings along the route with points so that the system can speak their presence when a traveler is passing by.

Also there is a concept of path synchronization points, something on the path that could be marked out for a traveler to touch or here to assure them that they are in synch with where the system says they are. GPS signals do not work in buildings, or under covered walks so this could also help navigating a blind person when the GPS signals are not available, although here the synch points could not be associated with a point but with a path position along a route.

Since there is no GPS within buildings, a route description would have to suffice, although this route could be made up of appropriate synch points, with descriptions on how to get to the next point and so on until the desired destination was reached. This type of route could be stepped through in segments at the users pace using a forward, backward type of input. The compass will still work here and be helpful.

The compass can be used to help the traveler query the system, for instance if it is mounted on the head, say on the earphones the traveler could "look" in a direction and ask what is in the direction I am looking. Even still we could use a level and ask what is down to my left.

The information should be accessible to all that can use it. It will be growing as new routes are added; also some routes may be altered due to the constant construction on campus. Therefore this information should be accessible over the campus network.

Also every building on campus should eventually have its layout (entrances, and basic organization) descriptions in the system.

Can we construct the system in such a way as to have it discover a route for you from your current location to a system known location, maybe you happen to be on known path, or are close to one.

Maybe we could map in all basic sidewalk grids and work off this.

Do we include descriptions of walkway material?

I would like to make this information, layered so as to make it usable for people in wheelchairs,

People new to campus, or people touring the campus.

It would be nice to associate common traffic levels with routes as a function of time of day, day of the week, based on prior observations. For instance on weekdays while classes are being held from 12:15 p.m. till 2:00 p.m. the walkway between Turlington Hall and the CISE building is very crowded.