音声ブラウザご使用の方向け: SKIP NAVI GOTO NAVI

An Ultrasonic Mobility Device with Minimal Audio Feedback

P. Blenkhorn, S. Pettit and D. Gareth Evans
Technology for Disabled People Unit (TDPU)
Computer Systems Design Group
Department of Computation
UMIST
PO Box 88
Manchester
M60 1QD
United Kingdom
Email, P.L.Blenkhorn@umist.ac.uk
Tel: +44-161-200-3368, Fax: +44-161-200-3373

Web Posted on: November 22, 1997


Abstract

This paper describes an ultrasonic mobility aid that uses minimal auditory feedback. The paper discusses its operation and design and compares it with other approaches.


1. Introduction

There have been a number of systems that have used ultrasound in an attempt to support the independent travel of blind people. In simple terms, such systems generally operate by sending out a pulse of ultrasound. Eventually the pulse is reflected from a solid object in the path of the pulse. The time between the outgoing pulse being transmitted and its echo being received corresponds to the distance between the transmitter and the object. This information is then relayed to the user in some audible way. Most practical systems are rather more complex. They use multiple transmitters to derive a more detailed view of their environment. Two well-established systems are the SonicGuide TM (1) and the Sonic Pathfinder (2). The former provides users with a large number of environmental cues encoded in audio information. The Sonic Pathfinder provides fewer audio cues with the goal of avoiding information overload. Both systems provide information on a regular basis. Despite a wide range of these systems being available the take up by the user community has been low. We speculate that there are three reasons for this. Firstly, the systems are expensive. Secondly, they generally produce continuous audible information that is distracting to the user and, thirdly, most systems provide complex information that requires considerable training and practise to use.

Alternative systems use tactile, rather than audio communication with their user (3). These remove the audible distraction, but are likely to be expensive and to require extensive training. They do appear to have been widely used.

This paper describes an alternative ultrasonic aid, called JiL, which provides minimal, but essential, information to the user. It attempts to minimise cost by being a very simple device. It minimises distraction by only providing the user with information when there is an imminent danger of a collision or when the user specifically requests it. It requires virtually no training.


2. Motivation for the UMIST System (JiL)

The motivation for the work described in this paper was taken from a blind colleague who complained that he was occasionally injured by obstacles at head and chest height, such as door frames and overhanging branches. Our objective was, therefore, to develop a cost effective device, that warned a blind pedestrian of obstacles.

The work was constrained in one very significant way, namely that the system could not be head mounted (for understandable cosmetic reasons) and would, therefore, have to be positioned on a user's chest or on their cane.


3. JiL as a Head Level Obstacle Detector

The ultrasonic mobility aids identified earlier provide significant amounts of auditory information to support orientation and mobility. They are generally used in conjunction with other aids such as canes and guide dog (4). JiL has a well defined single function, namely to give a user warning of an obstacle at head or chest level. This implies that, for much of its use, JiL is silent. Audible warnings are given only when an obstacle is first detected at a specified range.

JiL is designed to give warnings at two set ranges. Firstly, when an object is detected at the range of two metres, JiL sounds a tone for around half a second. When an object is detected at the range of one metre, JiL sounds another distinct tone, again for half a second. These are the only sounds that JiL produces in normal operation.

JiL is designed in this way to provide minimum feedback and hence distraction to the user. The precise distances at which JiL sound the alarms is configurable and evaluation of the system will attempt to determine the optimum distances for a range of users. This is closely related to the walking speed of the user. One goal is to examine how useful it is to be able to set the audible warning distance based upon the time taken to traverse that distance rather than a fixed length. For example, the warnings would then correspond to obstacles one or one half a second away. This is similar to an idea first proposed by Heyes (5). In addition, the pitch and duration of the warning tones will be decided by further evaluation.


4. JiL as a System For Exploring The Environment

The prototype form of JiL can also be used as a hand-held device that indicates the distance between a user and objects in the immediate environment. It operates in a similar manner to the Nottingham Obstacle Detector (6). JiL is held horizontally and the user depresses a button on the side of the device. For as long as the button is depressed, JiL will sound a continuous tone whose pitch is related to the distance between the user and the nearest 'visible' object; the higher the pitch, the nearer the object.

It is thought that this mode of operation will be useful in situations where navigation becomes complex, for instance in corridors in a suite of offices. The use of this mode will be investigated when JiL is fully evaluated.


5. Design

JiL has a number of significant constraints on its design. It must be small, light, cheap and have low power consumption. These are addressed by standard microelectronic system design techniques and the introduction of a small, low power microcontroller.

More significant constraints concern the design of the ultrasonic part of the system and the interpretation of its signals. Firstly, we use probably the simplest method of determining distance using ultrasonics. That is to transmit a pulse of ultrasound, and to receive its reflection from any objects in its path. The time taken for the reflection to be received is proportional to the object's distance from the transmitter. As sound travels at 332 metres/second, the reflection from the object one metre from the transmitter will be received approximately 6 milliseconds after transmission. Secondly, one must consider the circle encompassed by the ultrasonic pulse at a distance from the transmitter. Since we are seeking to protect the head and chest of a user and that the final warning signal is one metre away from the user, this signal must be well matched to the area of protection. Currently, at one metre, obstacles in a circle of diameter 0.7m will be detected. At two metres the diameter doubles to around 1.4m. This means that a number of 'two metre' warnings will be false alarms. Thirdly we must consider the attenuation of the reflected signal due to distance travelled, imperfect reflections and the fact that an object will not necessarily reflect the whole of the signal. Experiments show that amplification of around 24 dB is sufficient.


6. Results and Further Work

The current prototype version of JiL has been tested in laboratory conditions and is successful at its major purpose of identifying potential hazards, as small as 30 mm across, at ranges of one and two metres. We believe from limited environmental trails that the system will detect branches at head height. Further trials with real users are planned to further evaluate the device. In particular to experiment with gearing the device to user's walking speed.

Limited evaluation of JiL as system for exploring the environment has been undertaken. The system works well at distances of one to a few metres. As distances increase, the system is more susceptible to picking up false reflections and will miss quite large objects that do not fully reflect the whole of the ultrasound beam. Further evaluation is required to determine the usefulness of this feature.


7. Acknowledgements

The authors would like to acknowledge the funding of the TDPU by the UK Guide Dogs for the Blind Association, which has made the development of this device possible.


8. References

(1) L. Kay, "Evaluation of the Ultrasonic Binaural Sensory Aid for the Blind", Evaluation of Sensory Aids for the Visual Handicapped, National Academy of Sciences, 1971

(2) A. D. Heyes, "Sonic Pathfinder", Wireless World, April, 1984

(3) J. A. Brabyn, C. C. Collins and L. Kay, "A Wide bandwidth CTFM Scanning Sonar with Tactile and Acoustic Display for Persons with Impaired Vision (Blind, Divers, etc.).", Ultrasonics Int., pp 348-353, 1981

(4) L. Kay, "The SonicGuide TM, Long Cane and Dog Guide: Their Computability", Visual Impairment and Blindness, 75(7), 1980

(5) A. D. Heyes, "Microprocessor Techniques Applied to Ultrasonic Pulse/Echo Travel for the Blind", In Electronic Spatial Sensing for the Blind, Ed. Warren and Strelow, Dordrech; Lancaster :Nijdoff, 1985

(6) A. G. Dodds, J. D. Armstrong, C. A. Shingledecker, "The Nottingham Obstacle Detector: Development and Evaluation", Visual Impairment and Blindness, 75(5), 1981.