One major design decision on a technical level regards the degree of instrumentation of the smart object itself and its environment. This is not a particular issue of anthropomorphic smart objects but basically of all applications that combine smart objects and intelligent en- vironments. In this section we will discuss the possibilities and implications of embedding technology inside the object itself and shifting it to the environment, specifically for proto- typing anthropomorphic smart objects.
For the purposes of this work we distinguish four different classes of smart object imple- mentations, which are also visualised in Figure 4.2:
1. Self contained: All instrumentation resides inside the object.
A self contained smart object is able to perform all interaction methods and services without any other support of the environment or other infrastructural elements besides electricity supply, either continuously or through occasional charges. By integrating sensors, actors, computation and communication capabilities within the device, the re- quirements of cyber-physical systems as introduced in Section 2.1.1 are fulfilled as
4.2. INSTRUMENTING OBJECTS AND ENVIRONMENTS 103
Smart Object
Smart Object
Smart Object
Smart Object
Environment Environment Environment Environment
Self Contained Environmental support Depending on Environment No Instrumentation
High Degree of Natural Embodiment High Reuse of Technology
Figure 4.2: Different classes of integrating technology in smart object systems and instru- mented environments
well. Obviously, a major advantage is the autonomy of the device and thereby the possibility to easily transfer the system to any other location. The mobility criterion is much more critical for products and real-life applications than for pure prototypical development in laboratory environments or similar scenarios. However, the integration of all necessary hardware poses strong constraints on the overall physical construction of the prototype, depending on the application and its requirements. Each component requires space and adds weight, and many sensors and actuators require careful place- ment to accomplish their task. But at the same time, such a solution supports an object centric interaction model and yields full embodiment according to Fishkin’s taxonomy (see Section 2.3.1).
2. Instrumented with environmental support: Primary technology, which is required for main functionality, is embedded into the object, the environment provides addi- tional services.
Instrumented smart objects with environmental support as well are able to function independently retaining their basic functionality. But, in addition to that, such objects are also able to communicate with components of the environment, which possibly support the object with additional services. For example, cameras might identify the interacting user or loudspeakers render audio output. The main advantage of this ap- proach is the possibility to utilise theoretically unlimited resources of the surrounding
infrastructure and thus realising otherwise unfeasible functionality. It might also be advantageous to outsource hardware components and/or functionality to the environ- ment in order to save space, reduce the object’s complexity, and to make externalised components reusable for other smart objects. Of course the mobility advantage dimin- ishes with increasing dependency on surrounding infrastructure. Regarding the mental model of an object centric interaction, the developer has to consider the possibility that shifting sensors, computation or actuators outside the object may alter the users per- ception and conflict with the goal of achieving a high degree of embodiment. But this issue also depends on the employed modality and how contents are delivered. For ex- ample, auditory output could in many cases still be perceived as emitted by the device if rendered appropriately, e.g. through spatial audio (see more in Section 5.3.3). The result on the scale of embodiment would most likely be nearby or environmental. 3. Partially instrumented with main technology in the environment: The object is
equipped with a limited amount of instrumentation (e.g. RFID tag) and the environ- ment contains further required technology (RFID reader, sensors).
A partially instrumented smart object that relies on the availability of certain infrastruc- tural technology goes one step further and necessitates a specific environment in order to function as intended. An example would be an object that relies on an external com- puting machine to process and interpret sensor data, required for the interaction pro- cess. The apparent disadvantage of the dependency on technological infrastructure is a trade-off to the pragmatics of a centralised solution that simplifies development efforts of a multitude of prototypes e.g. due to fast changing requirements. Limited hardware resources can be efficiently utilised, shared and reused. On the other hand, choosing this approach foregoes obtaining insights on the technical feasibility of a higher level of embedment in real-world scenarios. Apart from that, certain sensory data requires to be collected within the object itself, e.g. for tangible input detection. Analogical to the previous class of objects, sustaining the perception of a self-contained and highly embodied object as the main interaction partner in this case is more difficult and needs to be addressed by other means.
4. No instrumentation: All technological components reside in the environment. At the other end of the scale we have smart objects that do not make use of embedded technology at all, but instead are part of an instrumented environment that hosts all application logic, sensors and display devices involved in the application. This is the case for instance when cameras are used to identify and track an object, while external displays are responsible for the system output. This situation exposes the developer to several problems:
• Interaction recognition cannot make use of integrated sensors (e.g. for accelera-
tion)
• A strong dependency on a very particular instrumented environment exists • Creating an impression of high embodiment is more difficult. More likely, the