| |
5.2. The Sensor-Opportunities-Based Approach
63
approach to smart objects would always require one central black-board compo-
nent, which is not ideal for distributed scenarios. The Context Information Service
described in
[
Pascoe 1998
]
takes an approach well known from object-oriented pro-
gramming. Artifacts are described as entities with states representing their context.
Further artifacts can have relations among each other. Using these relations a graph
representation for reasoning can be built up.
The work presented in this chapter also follows an artifact-based approach. Artifacts
represent real world objects with attached sensors and span the range from low level
sensors to high level abstractions. The notion of a setting-layer combines artifacts
according to their scenario of use into more complex, distributed and networked
perception systems. One central issue is collective perception, where distribution
issues directly related to sensing and perception are considered.
Sensor networks have been assessed as means for data collection in mobile and ad-
hoc settings in di erent research projects recently
[
Estrin et al. 2001
]
. Applications
span a wide range, but are most often related to monitoring particular parameters
in the environment
[
Estrin et al. 2002
]
.
In contrast to simply collecting sensed data from the environment, ubiquitous com-
puting follows the goal of enabling new applications by enhancing the environment
with communication and sensing devices. Even-though sensing is inherent in this
setting it di ers from the tasks faced in a classic sensor network approach. When
building applications the meaning of sensor data emerges from the coupling of the
sensor with the object it is placed on. E.g. the desired temperature of a co ee cup
and that of a ice-cream bowl di er substantially. To address this coupling between
sensors and artifacts we present an approach directly tying perception to artifacts.
5.2 The Sensor-Opportunities-Based Approach
Today’s technology o ers a wide variety of sensors. Although many sensing applica-
tions have been produced, there is no support for the design of applications o ering
physical interaction. In order to make a step towards such a design framework
this section analyzes di erent means of sensing of humans and human activity. In
particular we identify six sensing goals, referred to dimensions of sensing: ID (1),
Object Use (2), Location (3), Bio Signs/Emotions (4), Activity (5), and Interaction
Among Humans (6). Those dimensions together with di erent sensor placements
are used to review and analyze ubiquitous computing research related to physical
interaction and sensing. The final discussion draws conclusions from this analysis
with respect to the appropriateness of sensors and sensor placement for di erent
sensing dimensions.
|  |
|
| |
|
|
