The CORES Project

P.A. Cudd+, M. Freeman+, R.B. Yates+, A. J. Wilson+, M.P. Cooke+, M.S. Hawley*
+ University of Sheffield, UK * Barnsley District General Hospital, UK
Abstract. Discussion paper of a proposed project that is the initial phase of a far
reaching research programme. The CORES project aims to investigate the
development of a speech-only-input based Human Computer Interface 0t(2I)
which will lead to suitably skilled disabled users being able to operate any
commonly available commercial personal computer or workstation. The long term
aim is to generalisc this to a portable device with an adaptable HCI (both in
hardware and software terms) suitable for disabled and able-bodied people alike.

1 Introduction
The UK lags North America and Scandinavia in establishing multi-disciplinary
research and development centres in the fields of Rehabilitation and Human-
Machine Interfacing. There is a direct correlation between the higher levels of this
type of research and the degree of productive employment and social integration of
disabled people in these geographical regions. It is of note that the industrial sector
has also recognised that applications of technology for disability has mass market
applications among the general population 1.
The technology required to achieve rehabilitation becomes more complex when
those with severe physical disabilities are considered2'3'4. For such individuals a
very limited number of controllable movements are available to operate devices
which replace the motor and communication functions. Current assistive devices
clearly benefit disabled people, but they do not provide communication or motor
function at the speed or response that is required for the activities of daily living.
Electronic and computer technology has a major role to play in the process of
optimising the control of devices to substitute impaired function, both in its ability to
control flexible displays for presentation of options and control information to the
user, and, to process historical sequences of actions to predict those which are most
likely to be selected next. The potential of Automatic Speech Recognition to enhance
the HMI is great, but many issues have still not been fully investigated, e.g. safety for
control applications; automatic recognition for impaired speech; and tailored
Human-Computer Interfacing, in terms of both software and hardware2.
The CORES project aims to go beyond current investigations and initiate a five
to ten year research programme. The name 'CORES' is derived from
'COmmunication and Rehabilitation Engineering at Sheffield'. The project will
contribute to the long term goal to develop a Universal Human-Machine Interface
(UHMI), i.e. a portable device that would facilitate the operation of any type of
computer or industrial/domestic electronic control system by either able-bodied or
disabled individuals. Thus the concept goes beyond existing research2,3,4.
Many disabled people have both limited and uncontrolled limb movements while
still having some speech, but that may also be severely impaired. Nonetheless,
speech offers a useful alternative to other input modalities. Thus the CORES project
will aim to result in a design framework for the UHMI and a demonstrator with
impaired-speech input that will facilitate the use of commercial computers.
 494

2 The Universal H u m a n Machine Interface

For more than a decade the cognitive processes associated with human tasks have
been analysed to improve the effectiveness of the HCI. Through fundamental
research, this branch of systems development has now established a well founded
theoretical framework in which computer interfaces can be designed.
Unfortunately, little of this work has either been applied to the development of
generic electronic systems for disabled users or to include consideration of the
optimum hardware. So currently commercially available state of the art HCIs such
as Microsoft Windows TM and X-Windows" require an input morality transducer for
non-standard input devices. Interaction (i.e. control of) computers by disabled users
is an area that has received some attention. However, the HCI aspects of the existing
systems to be used, i.e. the standard hardware input devices and the operational
characteristics of the software application, have been accepted as a starting
specification. So, problems, or at least awkwardness of use, appear because the input
device and the feedback needs of the user were not part of the original HCI. The
CORES project demonstrator system aims to provide a tailored interface for the
chosen input modality (speech) which maximises independence of the application-
user interface. Implementation of the demonstrator based on user- and application-
adaptable state of the art hardware gives the CORES project a significant edge over
previous approaches to HMI for the disabled. The work of CORES therefore points to
the possibility in the future of bespoke user interfaces for everyone; in the more
immediate future this may only be justifiable for specialist applications.
In industrial and business applications increasing amounts of computer system
processing power is devoted to the user interface, viz. Graphical User Interfaces,
multi-media, speech recognition and virtual reality. Current hardware
implementations are not tailored to these input/output (I/O) tasks, and consequently
the implementations of even well- researched and designed HMIs are frequently
slow. When such interfaces are considered for use by the disabled community, they
are not sufficiently adaptable to cope with the extra-ordinary needs. These factors
have led to the recognition of the need for a new approach in designing a Human
Machine Interface (HM 0. Future implementations of HMIs should be portable and
be conceived as a quite separate I/O device to the 'application tool'. The latter
'application tool' could be a computer or an industrial/domestic electronic system. Of
particular interest, therefore, is a framework for the development and evaluation of a
modular and integrated Universal Human Machine Interface (UHMI). This would
include software, operating system and hardware, and be modular in terms of its I/O
capabilities, these modules being tailorable to the appropriate requirements of the
user and the attached 'application tool'. Such a framework would bring together the
host application (e.g. word processor) with components of communications
technology (e.g. speech recogniser) and the user-application interaction.

3 W o r k to be carried out
The first task is to create a requirements specification for the framework, using
information from related work such as in the area of telecoms5. Specification of
 495

hardware for the demonstrator will also be initiated. Next, the framework will be
developed using available communications technology resulting in a demonstrator.
Preliminary evaluations would be carried out and reported. Lastly, investigation of
more specific day to day and long term needs of disabled individuals would be
addressed, e.g. casual communication, and, facilitating education and employment.
It is difficult at to give more precise details on the work to be carried out until the
requirements specification has been achieved, but we expect to consider the
following themes taking full advantage of existing HMI tools and experience:
9 Existing speeeh recognisers and impaired speech
9 The parameters of impaired speech
9 Gesture recognition based on recent advances in machine vision
9 Input systems based on movement analysis
9 Actuator feedback and cognitive issues
9 Use of parallel input and output modalities
9 Adaptive HMI
9 Commonality with applications using processing intensive user interfacing.
It is envisaged that the user will effect control through supplying input to a
device which 'connects' this input to the computer system in such a way that
manipulation of the processes on the computer is possible. It is planned that
computer platform independence will be achieved through use of those input/output
connections that change little with time, e.g. mouse, keyboard and display ports. It is
intended to use recent advances in 2D object recognition and ASICs to achieve
automatic recognition of screen activity.

4 Conclusions
Commonality with other processing intensive user-interfacing applications, such
as Virtual Reality, means that the future holds a greater opportunity to achieve more
powerful, bespoke and hence more appropriate user interfaces for disabled
individuals. The complexity of augmenting or assisting impaired functions will, by
necessity, mean that the approach to the technology required to realise user-friendly
and use-efficient systems be both complex and innovative. Development of bespoke
user interfaces in the Rehabilitation field has useful implications and indeed may
lead the way for the general Human-computer/machine-interfacing of the future.

Acknowledgement : The authors are indebited to Dr. P. Courtney.

References
1. The Apple Macintosh and Disability Access Syllabus, 1994.
2. Hawley, M.S., Cudd, P.A. and Cherry, A.D. : 'Implementation of a PC-based
integrated control system for children', Med. Eng. Phys., May 1994, V1.16, 237-42.
3. Buhler, C. : 'Uniform user interface for communication and control', Proceedings
ECART2, Stockholm, May 1993, pp22.3.
4. van Woerden, J.A. : 'M3S A General Purpose Interface for the Rehabilitation
Environment', Proceedings ECART2, Stockholm, May 1993, pp22.1.
5. yon Tetzchner, S. (Editor) : 'Issues in Telecommunication and Disability', CEC,
EUR 13845 EN, ISBN 92-826-3128-1, 1991