POSITIONING SYSTEMS FOR MOBILITY OF THE BLIND, OVERVIEW AND ANALYSIS

1. INTRODUCTION

Multimedia brings a whole new universe of applications for improvement of one's Quality-of-life, and this is especially true for the blind. Emerging multimedia standards suggest wide acceptance of new technologies by the blind users, both custom made and off the shelf, and with intention of promoting their active role in society. Unassisted mobility thus becomes an important issue with increasingly dynamic population of the blind.
Absolute positioning systems give position of a user on a map, using fixed points in space as reference. Mobile networks bring a plethora of location-based services to their users, but with limited spatial accuracy. Satellite positioning systems are inherently more precise, but they suffer from shadowing problems, and uncertainties related to motion.
Relative positioning systems are aids used by the blind that are referenced to the user him/herself, and these include all short-range environment-sensing devices such as sonars [1], radars, lidars, and soon perhaps video guidance systems based on computer vision. Unlike absolute positioning systems, precision of these is improving as distance to user decreases.
Goal of this paper is determining equilibrium between the two for the benefit of the blind users. Suitable interface is essential for use of integrated combined systems.

2. TECHNOLOGIES FOR THE BLIND AND QUALITY OF LIFE ISSUES

Sight is a valuable sense, a source of information from the world outside. 83 per cent of all messages that arrive to a cerebral cortex are sight-related [2]. For this reason persons with defective sight (blindness or low vision) have specific hindrances functioning in a world increasingly dedicated to seeing.
New technologies dominantly employ visual information, thus making performance of daily activities more efficient for people with healthy eyesight, but there is also a need for accessible technologies dedicated for people with sight deficiencies.
Visually challenged children in early childhood use aids in form of auditory toys (rattles, balls, etc.), and tactile toys. At school age, materials written in Braille script are of great importance.
Enabling computers to "speak" makes information equally accessible to the blind and seeing users, especially if they can not read in Braille. This reflected in dramatic improvement of Quality of life, both to children and to adult blind users, and it enabled them employment in new and more diverse occupations. This issue is of special interest to people who lose their vision at later age, and who already use new technologies in their work and everyday living. They have to migrate to a completely new way of using the same technology.
The same principles apply to mobility, and necessary skills are taught by an expert mobility trainer (peripathologist). These skills include unassisted mobility by means of using a (long) white cane, a guide dog, or some electronic mobility aid.
Work and mobility are the most important fields of life aided by assistive technologies, but there are many more, e.g. household and leisure. These aids are often built in a way that seeing users may use them as well, and in a way they are used to, thus promoting integration of accessibility into everyday appliances. It is important to emphasise that functionality of these aids is not fundamentally different, but the prices are usually several times higher than the equivalent products for the seeing population.

Due to the widespread availability of mobile networks, it is worth examining their potential applications concerning mobility of the blind. Most common system, with number of users approaching one billion worldwide is GSM, and there are several methods already implemented with it [3]. These methods are listed as follows:

Cell type information gives information about a serving cell. This is usually too coarse for any positioning, except for dense urban environments with closely spaced picocells.

Positioning by time advance (TA) is based on GSM system's method of compensating for signal delays. Unit delay corresponds to one bit, its duration is 3.70 , and it gives a TA unit distance of 550 m. This is also too coarse, and some improvements are needed.

TOA and TDOA measurements are approaches to improving TA information. TOA, time of arrival, determines a position of a user by triangulating TA data from at least three base stations, while TDOA, time difference of arrival, observes their differences.

Positioning based on received signal strength measurements is frequently debated method, but with implementation problems due to complexity and precision issues. It is based on erroneous assumption that signal strength is uniformly decreasing with distance, which is not true for modern mobile networks with high gain antennas. This is illustrated by Fig. 1. Accuracy is also dependant upon mapping accuracy (usually about 25 m), and predictions' standard deviation, which is between 6 dB and 10 dB [4]. As this approach requires long-term averages and history of movement, users' privacy issues can not be neglected.

Fig. 1 Signal profile mask for typical mobile network cell with 15 dBi antenna

UMTS and 3G systems in general provide their services at data rates much higher than GSM. Location based services based on TA in UMTS will have precision of about 80 meters, which corresponds to chip rate. This will suffice for all commercial applications, and may be considered useful for mobility applications for the blind.
GPS system accuracy depends mostly on local environment, and varies between 5 and 10 m [5]. It is well adjusted with electronic map accuracy, and may be considered as a reference system for absolute positioning. There are some constraints though, and these include high power consumption, large physical dimensions, and false readings in high urban environments due to shadowing and reflections. Movement direction estimation is dependent upon users' speed and accuracy of positioning.

Various sonars with purpose of improving mobility of the blind are available for several decades now. There are different approaches to solving problems of interfacing environment information to the blind user, but the most useful so far is a system developed by Leslie Kay [6], which employs binaural [7] sonification [8]. Binaural environment representation allows users brain to distinguish relevant objects similarly to the cocktail party effect, which is also related to binaural hearing, and it is best described as immersion into sonified surrounding.
Lidar, a device similar to radar in principle and operation but using infrared laser light instead of radio waves was implemented in various aids so far, most prominent of all is a Laser cane. However promising this technology may seem, there are many problems in its use, and all are related to optical properties of the environment. Mostly it is glass and metal, and problems are related to transparency or perfect reflection.
Radars were not employed for mobility of the blind yet, but UWB (ultra wideband) radio technology may change that very soon. Problems related to wearable radars are antenna size, interference with other radio systems, radio licences, and high bandwidth required for spatial resolution. All these problems are solved with this technology, and development of applications may be expected very soon.
All relative positioning systems are user referenced, and provide information about immediate surroundings. Useful range is short, in order of 10 meters, but much wider than the tactile range of a white cane. All of them suffer from misdetections of holes or voids in a path, and thus can not be considered as substitution for a white cane.

Integrated multimedia devices for guidance of the blind are comprised of several different and mutually independent devices. Various sonic interfaces are common in nowadays devices, and these come in form of signalling, sonifying, and synthetic speech [8]. Lack of privacy when using sonic interfaces is a problem worth noticing. Tactile interfaces are viable, but more expensive, and limited in information volume.
It is important to emphasise orthogonality of information sources as a requirement for successful interface of an integrated guidance system. Orthogonality may be achieved by means of discrimination, or sequentiality of data. Synthetic speech can be easily discriminated from binaural sonified environment, or from some simple sonic signalling. Using several sources of the same kind concurrently is confusing, and stressful.
Synthetic speech interfaces are essential for nowadays aids for the blind. While natural high quality synthetic speech is essential for reading applications, it is not optimal for navigation. Ambiguities in natural speech must be avoided by careful choice of messages conveyed to users. Orthogonality of speech information is maintained in orthogonal languages, and a solution comes in form of a "basic" constructed language [9].
Simplified language called "Basic English" was released in 1930 by its developer Charles K. Ogden [10], and it is comprised of 850 mutually orthogonal (orthological) words. Other basic languages developed so far include Danish, French, German, Russian, and Spanish, [11]. Development of fully orthogonal basic language for any natural language is a linguistic problem, similar to development of every constructed language. There are over 300 constructed languages in the world now.
Using basic language is mostly a one way process. Listening is easy and natural, while speaking requires a learning effort, a process completely avoided in case of machines using basic language to deliver information.

Mobility aids for the blind are becoming increasingly complex and useful. Their impact to Quality-of-life is obvious, and expressed in terms of unassisted mobility. Development of new and improved aids is ensured by increasing number of the blind with financial power, due to the effective ageing of overall population [12].
Rapid development of mobile networks results in various services that are useful in applications for mobility of the blind. The most interesting are location-based services, multimedia communications [13], and development of text to speech (TTS) synthesisers for automated responses. It may be expected that 3G location-based services will surpass GPS in non-professional applications, which includes mobility of the blind. Multimedia communications make remote guidance by the seeing operator possible [14], and TTS gradually becomes a core of many human interfaces in applications for the blind.

Mobility of the blind may be observed by sensing ranges:
- immediate surrounding - includes senses of temperature, wind, moist, noise, smell, etc.;
- white cane tactile range - up to one meter, essential for safe mobility;
- relative positioning aids - up to 10 meters, nonessential but useful for navigation;
- absolute positioning aids - important for recognising macro-environment.

The later two are nonessential navigation aids, but increasingly important for safe mobility. In classic mobility training, a blind person is encouraged to learn everyday routes with assistance of mobility trainer [15]. The main assumption in this training is constant availability of a single safe route. In cases when main routes are blocked, navigation aids become indispensable asset.
Relative and absolute positioning systems are similar in a sense of locating a user in his/her environment, and thus their information is not completely orthogonal. They are not rival systems, since both are needed for successful navigation.
It is obvious that integrated mobility systems for the blind are expanding in their complexity. Components of such a system may be diverse, and physically separated from each other. With such diversity of components and their preferred interfaces, it is necessary to maintain mutual orthogonality of information to the blind user, since non-orthogonal concurrent information is confusing and tiresome. Signalling, sonifying, and speech information are mutually orthogonal by discrimination, and for concurrent synthetic speech sources orthogonality is maintained by sequential appearance. Understanding of conveyed messages is further improved by using a simple orthological language.

[1] D. Virkes, K. Nenadic, V. Budis: Sonar as a device for unassisted mobility of the blind, KOREMA Automation in Transport 2002
[2] A. Kovac, Godisnjak Hrvatskog saveza slijepih za 2003, Hrvatski savez slijepih, 2004.
[3] N. Matosic, Mobile Station Positioning Methods, Proceedings, ELMAR-2002, Zadar, Croatia, June 2002
[4] D. Virkes, Propagation Modelling Using Slope Effective Height Algorithm, Proceedings, 16th ICECOM, Dubrovnik, Croatia, October 2001
[5] R. Filjar, Horizontal GPS error distributions in different positioning conditions: a case study, Proceedings, ELMAR 2002, Zadar, Croatia, June 2002
[6] L. Kay, Auditory perception and its relation to ultrasonic blind guidance aids, J. Brit. Inst. Radio Engrs, 1962; 24:309-317
[7] J. Sunier: Binaural In Depth, The Binaural Source, 1999, www.binaural.com
[8] R. W. Massof, Auditory Assistive Devices for the Blind, 2003 International Conference on Auditory Display, Boston, MA, USA, July 2003
[9] D. Zerec Virkes, Constructed languages - Basic and Special English, Final Thesis, University of Zagreb, Faculty of Philosophy, Department of English, September 2003
[10] C. K. Ogden, Basic English: A General Introduction with Rules and Grammar, Paul Treber & Co. Ltd., London 1930, 1940
[11] Ogden's Basic English, Web page, http://ogden.basic-english.org/basiceng.html
[12] WHO Fact Sheet N° 146, February 1997
[13] B. Zovko-Cihlar, Wireless multimedia networks, Proceedings, ELMAR-2002, Zadar, Croatia, June 2002
[14] Garaj, Cecelja, Balachandran, The Mobile System for Remote Sighted Guidance of Blind and Visually Disabled Pedestrians, Brunel University, www.brunel.ac.uk/faculty/tech/
/systems/research/ElectronicResearchGroup/GPS_Vania.html
[15] G. Zovko, Peripatologija I, Skolske novine, Zagreb, 1994.