Bristol Wearable Computing
The Bristol Strategy for Research in Wearable Computing
Broadly, the aim of the Bristol Wearable Computing Initiative is to answer the question: What if computers could be worn as unconsciously as clothing? We want to develop an inspiring and credible vision of wearable computers that will bring them into the mainstream. We view the wearable computer as the future personal computer - a computer with renewed focus on the personal, reflecting technology improvements over the 16 years since the first PC. Specifically we are interested in the wearable computer as a personal system supporting a range of applications, rather than as a personal appliance supporting just one, and as a personal buy for the individual rather than an organisational buy for, say, the office worker. Inspite of our lofty ambitions, our approach is to stay practical: developing promising wearable computing opportunities through implementation and collaboration.
In the next section we explore what is new about wearable computers and use this to develop a sense of their full potential. After this, we ground our theory using specific examples of the what can be done with wearables. This is presented as a speculative account of how these devices will develop from where we are today. Finally we indicate where we place our bets with our own research programme.
The potential of the wearable computer
The wearable computer is distinct from its predecessors in three principle ways: usability, situatedness and connectivity.
The wearable computer can be seen as a consequence of the inevitable improvement in technology that allows manufacturers to make smaller and lighter computers. In this sense wearable computers are about improved usability through the evolution of ergonomics. This overlooks the significance of unconscious wearability, which is both deceptively hard to achieve and at the same time radical in its implications for end-user value.
First, consider end-user value. Unconscious portability enables a switch in the role of the computer from being a planned-usage device to being an always present device. The planned-usage device is sometimes taken and other times left behind depending on whether the user is able to anticipate need at the moment of picking up the device. The always present device is taken everywhere without the user going through any planning process - it is taken even if no need is expected, because there is no cost in carrying the device and "you never know..". In fact the always present device is able to contribute extra value disproportionate to the number of extra hours for which it is carried. This extra value comes from the always present device being able to address three new types of task. The ability of wearables to address these kinds of task supports our claim that wearables are radically rather than merely incremently different from other types of portable device:
In addition, the always present device can be used without the disruptive conscious query: Have I got my computer? Freed from this disruption it is more possible that the user may come to operate the device subconciously. In many application areas the ideal is that the computer becomes "transparent", allowing the user to focus on the primary task.
Now, consider what is meant by an unconsciously wearable device. It looks straightforward, but in fact wearers may be conscious of their computer in many different ways. Here are the requirements for the ideal unconsciously wearable device:
There is a further and more subtle way in which the wearable computer is more usable than its predecessor and this is developed towards the end of the next section on situatedness.
The wearable computer has a tremendous opportunity to deliver value to wearers by monitoring their physical situation. Physical data about the wearer's situation can be valuable on the spot (as in the case of navigation systems) or it may serve the purpose of memory augmentation (recording the users experiences for later review). The challenges are:
Environmental data is useful for a whole range of applications. In the examples above the data is presented explicitly to the user. More generally, environmental data can also serve a behind-the-scenes role, controlling what and when the wearable does, but not necessarily ever being viewed directly by the user. This leads us to the broader area of situated computing. Situated computing concerns detecting, interpreting and responding to the user's local environment.
Situated computing uncovers a broad range of promising applications, each worth study in their own right. In addition it enables a new relationship between user and computer. Namely, situated computing allows the wearable computer to act as a mental extension of the user. Being a mental extension means that the computer operates from a shared mental context with the user. The wearable computer can do this by interpreting physical sensor data from the user's environment.
The interpretation problem can be near-impossible as in the case of the of deducing that the wearer is 'sad' from a series of biometrics or it can be as simple as deducing that the user is 'going shopping' because they are in a shop. In any case when a computer operates from a shared context with the user, human-computer interaction becomes much simpler and much more spontaneous. This is because traditionally much of human computer interaction is about setting the computer's operating context to match your own. In particular within a computer we formalise the different contexts that a user may operate from as 'directories'. The effort of arranging and traversing directories is all to do with reflecting different user contexts within the computer's world. From the user's point of view this is all "overhead" and therefore frustrating; the user is not concerned about setting up the computer in a "going shopping" context, they just want to see their computerised shopping list.
The desktop personal computer never had the capability to set its own context to match the user's, firstly because it was 'in the dark', having a very limited input channel from the real world, and secondly because it was stuck in one place and therefore disconnected from the user's physical context. The wearable computer fitted with physical sensors has the potential to overcome these difficulties and operate from a shared mental context. This presents us with a radically new opportunity to redress the challenge of making a truly personal computer, an "ultra-personal" computer.
Connectivity concerns the connection of the wearable computing platform to "context drivers" and IO peripherals (hot plug and play) and to local and global resources (sometimes connected, always in touch). These connectivity issues are driven by the wearable's situatedness: the wearable is no longer a computer sitting uninformed in a corner, it is now the wearer's full-time digital ambassador [err..pretentious?] and must establish connections with other digital things according to the user's situation.
Being connected to other things is not automatically useful, rather the wearable should establish connections that allow it to provide services uniformly across different situations. Let's look at what this means for drivers, IO preipherals, local resources and global resources:
Context drivers are software objects that allow the interpretation of physical sensor data. These drivers may be specific to sensor types, for example software that converts signals picked up from active badges into companion arrived events, or they may incorporate local knowledge, for example, an object that recognises the fire alarm sound and then uses location information to guide the wearer to the most appropriate fire-exit in the event of a fire. Dynamic installation of context drivers from local servers is the scalable way to make wearables artificially intelligent, the end result is that the wearable is able to exhibit a uniformly high level of local knowledge across different situations.
IO peripherals include physical sensors as well as user input and output devices. We could talk about the ideal wearable simultaneously combining all sensor types and all user IO types, but such a device could never approach unconscious portability. Instead a more pragmatic view is that the user will occasionally reconfigure their wearable by gathering a different set of peripherals with their base wearable unit. They will chose peripherals according to their personal preferences and what they know of the day ahead. This allows users to maintain a uniform quality of service relative to their needs. If this kind of "hot plug and play" is to be performed routinely then it had better not involve too many wires. One vision is that the wearable will be able to communicate wirelessly with peripherals such as displays, such that a user need only put down one display and pick up another for the wearable to redirect visual output appropriately.
Local resources comprise things such as printers, fax machines and desktop servers. Users may require services based on the "nearest printer". In order to support this type of service uniformly across different locations the wearable needs to maintain connections to the locally-available resources.
Uniform connectivity to global resources, such as the internet or remote colleagues, is desirable but poses some well-known communication technology challenges (eg using your mobile in some obscure part of Scotland). The objective is typically to provide the user with the sense that they are always in touch, inspite of only sometimes being connected. The wearable can help by using data caching and message buffering strategies to disguise the hard truth of intermittent connectivity.
So in conclusion the three principle distinctions between the wearable and its predecessors are:
Issues from this section are summarized in the diagram below: [click for full screen]
An account of the future of wearables
[How should I arrange this section and have I missed any better examples?]
What are our goals and non-goals for research in wearable computing?
Develop a vision
Our main goal is to develop an inspiring and credible vision of wearable computers. We will achieve this through selective implementation and collaboration. We will avoid developing enabling technologies unless justified by our main goal. This is appropriate for pre-competitive research with limited resources.
Focus on the horizontal market
We will focus on wearables for the horizontal market. HP is most likely to become involved in wearable computing starting from its presence in the desktop, notebook and palmtop computing markets.
Focus on the domestic market
We will focus on wearables for the domestic market. It is quite possible that HP may eventually make wearables for the business market, but in the absence of clear signals we prefer to address the home market, since here we are most clearly involved in a 'personal buy', which allows us to focus on delivering value to the wearer without the distractions of the demands of an MIS manager.
Explore user values
Inspite of our horizontal focus we are also concerned to understand the values that a horizontal wearable might deliver to users. In particular we are interested in understanding which applications will drive the seeding and subsequent rooting of wearable computers.
Investigate the wearable as a mental extension
We are committed to examining the potential of wearable computers as mental extensions of their users. This is ambitious, but the possibility is too exciting to be ignored.
Avoid sensor distraction
There are no 'best sensor' technologies that the BWCI might try to 'own'. The value of a sensor type derives from the applications that it supports. Nonetheless we need to develop a range of sensors in order to make our computer real. Our strategy is to utilise student projects to furnish our prototype wearable with sensors but not to let sensor technologies distract us from our main goal.
Make wearables practical
We will focus on making wearables practical for everyday use. We need to be careful in separating out areas where underlying technologies will become more wearable in the course of time without our intervention and those areas where work is required to make wearables wearable. In particular we want to explore more wearable alternatives to the eccentric, pioneering constructions developed at MIT. This may cost us some press-coverage!
Focus on dynamic connectivity
Dynamic connectivity is to be a major focus. It is crucial for wearables but also has relevance for disaggregating PCs, notebooks and palmtops. My only concern is that this area is in principle addressed by distributed computing theory. If this is the case then our work is primarily a proof of concept in the wearable domain, together with scenario planning that shows how sensory and information infrastructure for wearables may evolve on a commercial basis. For example, we need to explore how the web may provide the basis for information databases describing, for example, the location of all Pizzahut's UK outlets. We also need to examine the client-server balance for wearable computers in touch with servers.
Power supply issues
Power supplies for wearables. This is a difficult area to judge. Will wearable computers take off in the absence of special power solutions such as shoe-base actuators, inductive charging or solar panels? This depends on the early applications.
Our primary focus is to be on the socially acceptable wearable. Thereafter, privacy/monitoring and ethical concerns may follow. These will be solved by architectural means.
The wearable may be expected to synchronise data with other machines, for example, the user may have a separate desktop computer. I would expect this problem to have been resolved in the marketplace by the time wearable computers are a reality. Hence this is not an area for the BWCI.
ęCopyright Hewlett-Packard 1997, 1998.The material displayed is provided 'as is' and is subject to use restrictions. For problems or queries about this web contact web master. Last updated: 14 Jan 2000.