Home| Keynote Speeches| Workshop Papers| Other Documents
Summary
The MoBIC system is a prototype navigation aid for blind people which consists of a Pre-journey System and an Outdoor System. The Pre-journey System allows blind people to plan a journey by inspecting electronic maps using a computer with speech output. The Outdoor System consists of a portable computer which is able to use satellite positioning technology to calculate precise positions. The system is able to tell the user about their location when en route.
This paper overviews the project and specifically the six month field trial carried out in the UK. Some of the findings are considered in terms of (1) technical performance: does it work?, (2) real performance: can people use it?, and (3) reaction: do people like it?.
Finally, these results are considered in terms of how the MoBIC system might be further developed so its full potential can be realised.
Introduction and project overview
The MoBIC (Mobility for Blind and elderly people Interacting with Computers) project is a three year European funded project which aims to develop a navigation aid for blind people (see The MoBIC Consortium, 1997). The institutions involved in the project, and their key roles, are listed in the appendix. The navigation aid, the MoBIC System, combines a number of technologies: a computer with speech synthesiser, electronic maps, and the GPS (Global Positioning System, which uses satellites to locate precise positions). The resulting system has two parts: the Pre-journey System, and the Outdoor System.
The Aims of MoBIC
Stone (1995), like many authors in this field, makes a distinction between mobility and orientation. Mobility is the ability to move safely through the environment, avoiding obstacles and drops. Orientation is the ability to understand space and how you relate to it. Both skills are important if a person is to be able to independently move through the environment.
Mobility can be likened to low-level perceptual skills - the traveller is bombarded with close environmental information which he or she must make sense of in real-time. Blind people make use of primary mobility aids such as long canes and guide dogs to help them gather and interpret this information. For example, skilled use of the long cane will alert a blind person to drops of kerbs and steps, and obstacles such as lampposts and parked cars. Similarly, a well trained guide dog will stop at kerb edges, and guide the user around obstacles.
Orientation can be likened to high-level comprehension skills - a traveller makes decisions about where he or she is going, and the directions this will involve. Some of these decisions are made independently of information provided by the immediate environment. Sighted people can plan a route from my home to a destination without being able to see either. Of course a blind person can do the same.
The MoBIC project is particularly interested in helping blind people's orientation.
The Pre-journey System
When a blind person uses the Pre-journey System, it is analogous to a sighted person looking at a map in their home. Therefore, the system essentially gives blind people access to maps. The user can ‘explore' the map using cursor keys, and the system will describe the road, junctions, and directions through speech. Additionally, the map has information points which give useful details about the environment - e.g. house numbers, names of shops, locations of crossing points and bus stops, and the surface-type of the pavement.
The system is also able to calculate a route between two points; the user types two addresses, and the shortest route is calculated and described back to the user. This route can also be explored in greater detail.
The Outdoor System
Once a journey has been planned using the Pre-journey System the user may want to walk the route. In addition to their primary mobility aid (e.g. a long cane, or guide dog), the user can also use the Outdoor System.
The Outdoor System consists of a small computer powered by a battery pack. These are housed in a rucksack or waist coat (this is presented in figure 1). Also a GPS antenna is positioned upon the user's shoulder. This antenna is able to pick up signals from satellites orbiting the earth, and from this the system calculates a raw longitude-latitude position. The position is compared to the electronic map, and the relevant information spoken to the user through small headphones (the headphones used do not obscure environmental sounds).
The user issues commands through a small wrist mounted keypad. For example, by pressing the ‘Where am I?' key, details of the surrounding environment are given; or by pressing the ‘Next route instruction?' key a description of the next section of the planned route is given (e.g. "Walk east for 30 meters and then you will reach the junction of ..."). Importantly, the Outdoor System is designed to complement a person's primary mobility aid. It does not replace a person's long cane or guide dog.
In fact the Pre-journey and Outdoor systems are made up of the same components. The Pre-journey system is housed in a rucksack or waist coat, and is converted to the Outdoor system by simply swapping the Qwerty keyboard for the small wrist mounted keyboard, and adding the earphones.
Figure 1. A diagram showing a MoBIC ruck sack.
Six month field trial: overview of methods
The purpose of the field trial was to provide feedback for further developments of the MoBIC prototype (or any other development of similar products). Therefore, the field trial aimed to establish the performance of the system in an engineering sense, as well how users of the system were able to make use of the system to plan and make journeys. Additionally, we were interested in participants' reactions to the system, e.g. could they imagine themselves or others using such a system in the future?
The field trial lasted six months (March to September, 1996), involved 13 blind participants, and took place in Birmingham, UK. These participants were members of one of two cohorts (for the first three months of the trial, or for the second three months). A single participant was involved for the entire six month trial period. A complete description of the field trial procedure and results are reported in Bozic et al (1996), and Petrie and Douglas (1996), and an overview presented in The MoBIC Consortium (1997).
In this paper we present some evidence generated from the field trial to answer three key overlapping questions:
(1) Technical performance: Does it work?
The general functionality of the system is considered. This includes the consideration of six walks over a test route (by a sighted member of the research team). This analysis examines how well the devices were working.
(2) Real performance: Can people use it?
Participants were visited for training sessions once a week. The content of these sessions was recorded on session record sheets. These were then written up in the form of a diary, forming a case history of each participant's involvement in the trial. This analysis examines the success with which participants were able to use MoBIC.
(3) Reaction: Do people like it?
Two focus group meetings took place (one for each cohort), to discuss the ‘strengths and weaknesses of the MoBIC system'. These focus group meetings were recorded on audio tape and transcribed. This analysis examines how positively participants reacted to MoBIC.
These three different types
of data (system performance, observation notes, and transcripts of a discussion)
provide very different perspectives in answering the overlapping questions.
Technical performance: Does
it work?
The technical performance
of the system was obviously scrutinised carefully during the trial, and
inevitably any prototype system has technical problems. In the case of
the Pre-journey System the performance was recorded by trainers. In the
case of the Outdoor System, the performance appeared to vary according
to the environment in which the system was used. To establish a clearer
understanding of this performance some standardised tests were carried
out.
(1) The Pre-journey System
As described, the Pre-journey
System was designed to have a number of functions. This includes an exploration
function which participants could use to move a cursor through a map of
the environment. As the cursor moves features of the environment in that
area are described via a speech synthesiser - e.g. a bus stop, a pelican
crossing, a shop, a house number. These functions largely worked well.
However, the system found it difficult to interpret and describe complex
junctions such as roundabouts; therefore, descriptions presented to the
user were poor. In fact, during the trial, tactile diagrams were sometimes
used to help the participants understand these junctions.
While a huge amount of information was available on the map there were inevitably some errors and omissions. During the trial the research team were able to make corrections to the map when they were necessary. However, the setting up of the map data, and its maintenance, is an important issue which must be addressed if there is to be more widespread use of this sort of technology.
Another important function of the Pre-journey System is route planning. The route searching function calculates the shortest route between two points. Routes could also be ‘multi-legged', thus a route could be created which went from A to B via C and D. Similarly a route could be calculated which avoided particular segments of road of the user's choice. While this function is very flexible, there were certain aspects of a route which the program was not able to consider. Crucially the route planner could not account for the side of the road the participant was on, and was not able to make use of crossing points.
Another difficulty was that the map did not have some sophisticated database functions. While it is possible to search from one place to another, it is not possible to refer to these destinations in any other way than using the address. Therefore establishing a destination whose address is unknown was impossible.
(2) The Outdoor System
Undoubtedly, the Outdoor System functioned very well at times during the trial. It was able to pin-point users' position and inform them of it very accurately. However, during the trial a number of uncertainties regarding the accuracy and reliability of the Outdoor System arose. In some conditions the system appeared unable to give an accurate position. In order to clarify the circumstances in which this happened a member of the research team walked a circular test route (approx. 3km) six times using the Outdoor System. The route passed through a mixed environment including a shopping street, under trees, terraced houses with front gardens, and roads with different widths.
As the journey was made the researcher made notes of the accuracy of the signal. Additionally, the Outdoor System generated a log file of the positions it calculated.
Two maps are presented which contrast the route which was walked and the route which the Outdoor System logged (figures 1 and 2).
Figure 2 - A map showing the test route walked (including the side of the road)
Figure 3 - A map showing a route (9th August, 1996) calculated by the Outdoor System. The accuracy is generally high, though there are areas where the position calculated is inaccurate (shown as a broken line) and this is discussed in the text.
The most striking finding from this particular study is that on this route satellites are generally available, and the signal is often accurate (marked A on figure 3). The signal is typically on the correct road, and usually (though not always) at the correct location.
The findings also clearly demonstrated that location can have a major effect upon satellite availability and position accuracy. It appears that buildings in close proximity to the antenna have negative effects upon accuracy and availability. This was particularly apparent on a typical UK shopping street on the test route (marked B on figure 3). Although the street is wide, the buildings are tall and a pedestrian is typically close to the shop fronts. This resulted in (1) a drop in the availability of satellites, and (2) a drop in position accuracy.
This was backed up by the researcher's notes who observed a clear drop in accuracy on this road, e.g. the system reported a road junction 40 metres late. However, while accuracy dropped in this environment, the signal typically was on the correct road. Therefore, this information is still potentially useful to the blind user. However the system must be able differentiate accurate and inaccurate signals in order to inform the user of their current location in the appropriate way.
A further interesting indication from the study (although this is more speculative) is that the accuracy and availability appeared to improve in the more open environment of junctions. This may have implications for strategies of design and use of any future system.
Finally, on some of the test walks the initial signal was poor, and it appeared to take sometime for the accuracy to ‘settle' (marked C on figure 3). This may give the impression that the system is not working very well and the user may give up on some occasions.
In terms of generalising these results to blind travellers two cautions must be noted. Firstly, a typical blind user would walk more slowly than the researcher in the trials, and would occasionally pause (e.g. to interrogate the system, to post a letter, etc.). Therefore these walks do not reflect what would be typical use by a blind user. Secondly, the route walked is on a high part of Birmingham which may have increased the availability of satellites.
Real performance: Can people use it?
So far we have discussed some of the technical strengths and weaknesses of the system when it is tested in a functional way. However, the trial involved giving 13 blind participants a MoBIC system to use for an extended period of time. In the following two sections we consider what they were able to achieve using the system (from observations by their trainers), and how they reacted to the system (from focus group meetings).
A member of the research team met each participant once a week for a training session. The content of these sessions was recorded on record sheets which were eventually written up as case histories. Here we present extracts from these case histories as evidence of successful use of the MoBIC systems. Throughout this article the participants are identified by their initials.
Pre-journey System
The Pre-journey system allows a user to plan and explore routes from one place to another. There are many instances where participants (1) used the system to increase their spatial understanding of the environment, (2) found out about a route from one place to another, and (3) discovered the existence and location of particular amenities and services:
(1) Examples of participants using the Pre-journey system to increase spatial understanding of the environment
Participant TB used the compass directions and described forming an internal ‘bird's-eye' map of the matrix of roads. He was able to inspect junctions skilfully, learned to increase the step size and use this to estimate the length of roads.
PN noted that, "I didn't realise Station Road was a crescent." This road's beginning and end point are on the High Street, and PN described where these points are.
(2) Examples of participants using the Pre-journey system to find out about a route
The trainer planned a route from SC's house to a road where a friend of SC lived.. SC explored the complex route in detail, from which he seemed to have grasped the general shape of the route. SC was quite confident he could have walked the route if he were using the Outdoor System. If he were not using the Outdoor System he thought that using a small tape recorder would prove useful.
He and his trainer revised the exploration functions, exploring around PH's new home. The trainer then introduced PH to the planning part of the system, planning a route from his home to work. This took him a different way to the route he usually walks, but immediately PH learned something about the layout of the roads he had not previously known. This route was saved. He was very excited about the route planning functions, pointing out that this was something he otherwise just could not do alone.
(3) Examples of participants using the Pre-journey system to discover the existence and location of particular amenities and services
PN telephoned a local record
shop to ask for their address. He then keyed this into the Pre-journey
System and planned the route.
Outdoor System
The Outdoor System can give users information about their current position and the next section of their journey. For example, "You are at the crossing of High Street and South Street.", and "Continue north along High Street. After 30 metres you will reach the crossing of High Street and York Street." There are many examples of the Outdoor System being used to (1) help make decisions on the route, (2) alert participants when leaving a route, (3) reassure participants that they were on the route, and (4) as a memory aid.
(1) Examples of making decisions when on route
TB's general strategy was to confirm his position with the ‘Where am I?' key, and if this tallied with his understanding he would use the ‘What next?' key.
A journey was planned and walked from TB's house to the trainer's house - this passed past through a roundabout. TB entered, walked round the roundabout and exited up the correct road, pausing to confirm that the road was the correct one. If it had not been he would have tried the next road off the roundabout.
(2) Examples of alerted when leaving a route
PH walked past a side road which was on the opposite side of the road. The system warned him he had left the route. He crossed over the road and walked back, looking for the side road he assumed he must have missed. He found the side road. He also used his residual vision to check the road name on a sign and confirm it was the road he wanted.
PN had some difficulty remembering which way to go at a T-junction. He chose one way and was soon informed he was leaving the route, so, by a process of elimination, he turned around and went the other way. Here was a case when PN was able to problem solve en route.
(3) Examples of reassurance
PN made some hypotheses about the layout of roads, and then tested these by walking off route (overshooting a side road), having this confirmed by the system telling him he had gone off route, and then corrected his direction.
TB walked from his house to the trainer's house (approximately 1700 metres) with the trainer. TB had some prior knowledge of the area and had used the Pre-journey System to plan the route. It seemed TB had learned the route and was using the Outdoor System for confirmation.
One of the junctions on the route is a very gentle curve to the left. TB followed this curve round (correctly) without knowing. When the Outdoor system later told him where he was on the route he was able to interpret what had happened successfully. Without the Outdoor system TB would probably have become disorientated.
When TB was walking along straight roads, the descriptions of the environment broke the boredom; being told the distance to the next junction on the route was seen as excellent.
(4) Examples of memory aid
TB was very positive, commenting "I did not have to remember the roads like I usually do - like a sighted person. This is [the remembering] very difficult for long journeys."
Reaction: Do people like it?
Here we are concerned with
whether participants were positive towards MoBIC. This would give an indication
whether they could imagine this sort of technology being useful to them
or others in the future. The Focus Groups provided a context in which participants
could discuss these sorts of issues. Here we will discuss a number of issues
which were identified when this data was analysed:
· Positive reaction
to the system
· Technical aspects
· Safety
· Feeling conspicuous
Positive reaction to the system
A number of participants reflected positively about the current system and its potential in terms of the information it makes available:
"[It] has got amazing potential
... it is the equivalent of a sighted person looking at a map."
SC, Focus Group 1.
"I quite like the ability
to explore a map and that is where its strength is in a way. It gives you
a chance to look at an area."
PN, interview.
"I found the outdoor system
became like a sort of friend [..] it was company. It was very re-assuring
[..]".
TB, Focus Group 2.
Also participants considered the more general benefits which the system may make available:
"We went out together and
it was fun"
MH, Focus Group 2.
Technical aspects
Participants raised issues relating to the technical aspects of the system, which may be addressed in the future (in part at least). For example, the reliability and accuracy of outdoor system,
"This loss of satellite contact
is a problem that they have got to overcome."
KH, Focus Group 2.
and secondly the system size, weight and complexity,
"carting that around if you
were on a shopping expedition would be impossible I would think".
KH, Focus Group 2.
"[..] but I can imagine if
it was small enough to go into a pocket with ready answers and if you did
not lose contact with the satellite, it would be marvellous, I would take
it everywhere."
MH, Focus Group 2.
"I still haven't really been
able, having gone through the pre-journey system, actually myself to transfer
it from that to the outdoor system"
KH, Focus Group 2.
Although PH (Focus Group 2) called himself "a bit of a gadget freak", he found "it was still quite fiddly because it meant taking a couple of plugs out".
Safety
Safety was an interesting and complex issue. On one level safety was discussed in terms of the apparent inadequacies of the system (which may in time be corrected). For example, referring to the route planning function of the Pre-journey system, and the Outdoor System respectively:
"... often the shortest route
is not the safest route [because it can entail] crossing two or three very
busy roads which I could have avoided [..] knowing where the pelican crossings
were."
KH, Focus Group 2.
"in the middle of a crossing
[when] the system had lost the GPS, which is potentially lethal"
TB, Focus Group 2.
Contrast this with a general
discussion about safety. Both focus groups discussed the inherent dangers
in travelling as a blind person. Because there are always dangers involved,
it is logical to assume that a device such as MoBIC which can enable people
to walk unknown routes inevitably increases danger. This is a trade-off
between travel and safety, and what is considered acceptable safety will
vary from one person to another. The three examples below illustrate this
point.
Example One:
"BA: Well it is safe in
a sense that when you are walking on a pavement, but it does not tell you
if there is a skip there or a car coming out of a driveway. [..]
SC: It was not designed
to tell you about obstacles and things, that's still up to guide dogs and
canes etc."
Focus Group 1.
Example Two:
"SC: [..] but the outdoor
system says ‘you are now standing by a zebra crossing' and really you have
got to walk another 20 metres before you get to it. So you think you are
standing by a zebra crossing and you walk straight out into the middle
of the road, when you are 20 metres away from the crossing.
[..]
DV: The other thing is when
you are standing at a zebra crossing or where the traffic lights are, you
come up to [tactile markers] don't you?"
Focus Group 1.
Example Three
"Researcher: What about
the routes? The issue of the safety of routes.
[..]
PP: It depends on how deeply
you mean by safe. I was out on a new route for my test and the long cane
really came into its own then. It really did because I had no knowledge
of that route beforehand. For individual step by step the long can was
absolutely invaluable because it was a brand new route. [..].
TB: I suppose there is a
point there, there is static safety which could be programmed into the
machine, but there is safety from day to day and there is not a map in
the world that could help against that."
Focus Group 2.
In the first example, BA points out that there are dangers that the system does not predict, and SC replies that the system is not ‘designed' to help with these dangers. In the second example, SC points out the apparent inaccuracy of the signal may cause the danger of crossing a road in the wrong place. However, DV implies that other mobility skills can help you overcome that problem. Finally, PP and TB discuss that close mobility skills are important, and no system could alert the user to all dangers.
Feeling conspicuous
Participants in both focus groups described some of the ‘social effects' of the system. Some were positive, e.g. friends were interested (PH and SC). In contrast, there was also some discussion about how conspicuous the system might make you feel (which is reminiscent of young visually impaired people being reluctant to use their long cane or LVA):
"ER: I don't like personally
walking out with the MoBIC on with all these wires and everything.
SC: You're too vain!
[..]
SC: I do accept that the
rucksack looks a lot less [conspicuous] than that horrible waistcoat."
Focus Group 1.
Another reference to feeling conspicuous can be related back to the issue of safety:
"if the undesirable elements
in our society realise you have got at least a 486 on your back, they would
rip the chip off your back and leave you for gone"
TB, Focus Group 2.
Discussion and Conclusions
If we consider all the data presented in this paper it is possible to draw some conclusions about the potential of this prototype system.
Firstly, we presented many cases where participants had successfully used the system, and this is very encouraging. However, there were also cases where participants found using the system difficult. If we consider the most successful participants in the trial they all were familiar and confident with technology, and all were experienced travellers and had very good mobility skills.
Both these skills were important in the context of this trial. For example, some participants who were very competent with technology nevertheless found it difficult to extract any useful meaning from the Pre-journey System. They found the compass directions difficult to relate to reality, and felt that they were overloaded by references to roads, distances and directions. It would be hoped that through extended use of the device these skills would develop. However, this was not tested in the trial.
Possibly a consistently disappointing thing to come out of the case histories is that most participants rarely used the Outdoor or Pre-journey Systems when not with their trainer. Therefore there is a question related to their motivation to use MoBIC. So why didn't the participants make use of MoBIC more?
There are a number of potential reasons. Perhaps the most obvious, is simply related to some of the technical difficulties encountered during the trial. These are reflected in the section on technical performance above. These difficulties certainly put off some of the participants. For example, one participant tried a number of times to use the Outdoor System but had difficulties obtaining a signal, so gave up. Similarly, the reason may be related to the fact that it is not very easy to specify all destinations with the current system (that is, the system must have the street numbers of the starting and finishing points and can not cope if given information only of the sort "the supermarket").
Also, it is worth considering that the opportunity to interrogate a map and database of this kind is a new concept to the participants. Such an opportunity may mean a shift in the way they think about travel and mobility, and it may take a long time for such a shift to take place.
The participants were largely positive about their involvement in the project, and the potential of the system. Some reservations were related to the some of the technical aspects of the system (reliability of the signal and safety of the planned route). Also, there is great variance in the amount of information individuals require to make a journey with confidence. For some participants the information provided by the current MoBIC may be enough, for other participants major (perhaps unobtainable) improvements will have to be made.
An additional consideration is that of feeling conspicuous. Two contrasting examples of people's concern for the prototype system's appearance have been presented in the paper (not wanting to look different from others, and not wanting to have the system stolen). These issues must also be addressed in future development.
These issues have implications for the role of training in any future commercial system. It is likely that some users could be trained rapidly. However, other users would need a much longer period of training. In this case MoBIC would be one of many aspects of training which would support the development of the individual's general mobility and navigation.
A final consideration is a technical and economic one. Central to this technology are electronic maps. The electronic maps used in the Birmingham field trial were provided for free by Ordnance Survey (a UK map provider). If purchased commercially these maps would be prohibitively expensive. In the UK, the University of Birmingham, a number of national charities (RNIB, GDBA, St Dunstan's, and Blatchington Court Trust), and Ordnance Survey are currently discussing ways in which this can be overcome. Related to this, the map data used in the Birmingham field trial was amended by the research team with relevant information (e.g. location of bus stops, crossing points etc.). How this data collection and distribution is managed is an important question, and also a subject of our ongoing discussions.
References
Bozic, N., Douglas, G., Furner, S., Hill, E., Jansson, G., Johnson, V., Petrie, H., Tobin, M., Bringhammar, C., & Danielsson, A. (1996). Report on Birmingham field trial with recommendations. Deliverable No. 6.6. TIDE Project 1148 - MoBIC.
The MoBIC Consortium (1997). Mobility of Blind and Elderly People Interacting with Computers - Final Report. London: RNIB. (ISBN 1 86048 006 3)
Petrie, H. and Douglas, G. (1996). Internal Report on GPS availability. TIDE Project 1148 - MoBIC.
Stone, J. (1995). Mobility for Special Needs. Cassell: London.
The MoBIC project has a page
on the World Wide Web:
http://simsrv.cs.uni-magdeburg.de/~mobic/mobicuk.html
Appendix - Partners in the MoBIC project
Professor Thomas Strothotte,
UNIVERSITAT MAGDEBURG,
Institut fur Simulation
und Graphik,
Universitatsplatz 2,
D-39106 Magdeburg,
Germany.
Expertise: Software.
Mr. Jurgen Bornschein,
F.H. PAPENMEIER GMBH &
CO. KG,
Talweg 2,
Postfach 1620,
D-58211 Schwerte,
Germany.
Expertise: Hardware.
Dr John Gill,
ROYAL NATIONAL INSTITUTE
FOR THE BLIND,
224 Great Portland Street,
London W1N 6AA
England.
Expertise: User needs.
Dr Helen Petrie,
UNIVERSITY OF HERTFORDSHIRE,
Sensory Disabilities Research
Unit,
Psychology Division,
Hatfield,
Hertfordshire AL10 9AB
England.
Expertise: User Needs, Design,
Evaluation.
Professor Gunnar Jansson,
UPPSALA UNIVERSITY,
Department of Psychology,
Box 1854,
S-75148 Uppsala,
Sweden.
Expertise: User Needs, Design,
Evaluation.
Mr. Lars Reichert,
FREIE UNIVERSITAT BERLIN,
Institut fur Informatik,
Takustrasse 9,
D-14195 Berlin,
Germany.
Expertise: Software.
Mr. Stephen Furner,
BT LABS,
Human Factors Division,
MLB 2/8,
Martlesham Heath,
Ipswich IP5 7RE,
England.
Expertise: Communications
hardware, Evaluation.
Professor Michael J. Tobin,
Director,
Research Centre for the
Education of the Visually Handicapped,
University of Birmingham,
School of Education,
Edgbaston,
Birmingham B15 2TT
England.
Expertise: User needs, Evaluation,
Training.