CAPITULO II.............................................................................................................................. 2
3.3 RIESGOS
3.3.4 RIESGOS QUE SE DEBEN IDENTIFICAR
criterion like “the third entry” describes the referred object.
Cross-modal references - In a multimodal contribution, sometimes the content of one input modality refers to content that is provided by a second modality. For example, in the combination of a pointing gesture with the utterance “What is this building?”, the pronoun refers to the entity that is indicated by the pointing gesture. Ellipsis - In an elliptical construction, one or more words of an expression are omitted, e.g., if some of the constituents have already been mentioned in a previous turn. This especially may occur in information-seeking dialogues. The following example shows an elliptical construction:
User: What is the menu for today? System: (Presents the actual menu) User: And for tomorrow?
Constraints on References
Communicative acts that contain referring expressions can additionally provide restric- tions on the referred object which is valuable information for the resolution of matching referents. Pfleger introduced two types of constraints:
Syntactic Constraints - A referring expression can contain linguistic information about number, person, and gender of the referent. Usually these features must match the result of the reference resolution.
Semantic Constraints - A referring expression can also contain semantic infor- mation about the referent. For example in the utterance “turn on this lamp”, it is semantically clear that the user refers to a lamp and not to, e.g., a ventilator in a room. Semantic constraints can provide information about the type but also about features of an object like in the utterance “the green lamp”.
2.4 Cyber-physical Environments
The combination of the physical environment, the virtual world, and data from local networks and the internet is called CPE and enables a new spectrum of applications and business models. A CPE is the integration of several connected Cyber-physical Systems (CPSs) of the environment.
MacDougall (2013) defines CPSs as “enabling technologies which bring the virtual and physical worlds together to create a truly networked world in which intelligent objects communicate and interact with each other”. Thus, they integrate computation with physical processes. CPSs are a complement to embedded systems, which are engineered systems which combine computing with physical processes. Examples for the latter are
Figure 2.10 – A CPE is the integration of multiple CPSs (c.f. Kahl (2014))
automotive electronics, aircraft control systems, home appliances, etc. These systems all have one thing in common; they are realised in a closed box that does not connect its computing capability to the outer world. “A common feature of almost all CPS is that they heavily rely on networking” (Giese et al., 2011). Thus, together with the integration of modern multimodal human-computer interfaces and software-based internet services, this technology enables new interaction possibilities and applications that make the frontiers between the virtual and real world disappear. Furthermore, the connection of several embedded systems in the environment allows the provision of higher-level services.
Kahl (2014) gives an overview of the components in a CPS and the integration of multiple CPSs to a CPE. A CPS is a network of sensors, actuators, objects, and services (see Figure 2.10). Sensors perceive changes in the environment and can detect and monitor real objects and people. A set of dedicated services can process this sensor information and infer further findings about the actual physical context. Furthermore, a CPS is able to manipulate and interact with the environment by accessing actuators that form the counterparts of sensors. Then control commands in the form of electronic signals are transformed into physical measurable actions. The control of the actuators is a reaction to the processed sensor information. In a simple example, the signal of a motion sensor turns on the light in a room.
CPSs are an important requirement for the realisation of the Internet of Things (Wahlster, 2013). Together with the Internet of Services (Wahlster et al., 2014), they are key tech- nologies for the future projects Industrie 4.0 (Kagermann et al., 2013) and the Smart Service Welt (Acatech, 2014). Acatech (2011) see further potential of CPSs in energy (smart grid), networked mobility, health (tele-medicine and remote diagnosis), and in- dustry.
One possible application scenario of CPEs is the Smart Factory, where workers are supported during their work with sensor and service information, e.g., for localizing tools and construction units, controlling devices, or synchronising dates and tasks. The applied user interfaces are wearables or interfaces that support the interaction from a
2.4 Cyber-physical Environments 33
distance. Thus, the worker is able to retrieve information without leaving his workplace, dropping his tools, or polluting the input device.
2.4.1 Human Computer Interaction in CPEs
The emergence of CPEs requires a complete rethinking of the interaction between the environment and the users that become part of the environment, move in it, and interact with it. The classical interaction concepts where a user exclusively interacts with one stationary device, PC, or smartphone become less dominant and make room for concepts where the user interacts with the environment (Human-Environment Interaction (HEI)). Input and output devices are combined with sensors and actuators in the environment for the communication between humans and computers, and thus new interaction paradigms will evolve. Novel interface technologies like speech interaction, gesture recognition, distributed displays, eye-tracking, and wearable devices reinforce the trend to carry the interaction away from stationary places where keyboards, mouse, buttons, and switches are the usual input devices, into the room.
On the output side, the system may use classic devices for interaction, like displays or speakers. Additionally, new actuators are available in order to gain the attention of the user, e.g., a light spot that is turned on. On the input side, the stimulus for a system’s reaction can be triggered actively by the user, e.g., by pressing a button, or it can be an autonomous decision of the system based on received sensor information. However, if a user is aware of the trigger conditions for a system’s reaction, it is possible that he triggers a sensor event on purpose. Thus, the transition between direct and indirect interaction may blur. In the perception of the human, the old-established interaction with only one assigned modality for one functionality changes to an interaction with the environment independent from the actually involved sensors and actuators.
2.4.2 Requirements
Often, higher standards are demanded from an embedded system compared to general- purpose computing systems. From household and consumer electronics, a reliable and robust functionality is expected. A crashing stove or TV is not tolerated by the customer. This seems even more essential for aircraft or automotive control systems. Here, a malfunction can decide between life and death. In CPEs a large number of heterogeneous embedded and physical subsystems are networked and have to interact concurrently as well as collaboratively. Meanwhile, the environmental conditions can permanently change and the CPE must be able to adapt to unpredictable situations and be robust enough in order to react to subsystem failures (Lee, 2008).
Kahl (2014) argues that the communication infrastructure of a CPE must be flexible and preferably error resistant. Especially the number of integrated components is not a priori known and can dynamically change. He expects that an event-based communication
service like his Event Broadcasting Service (EBS) will be able to ad-hoc bind new sensors and actuators into a CPE. The concrete interactions can be seen as events at a specific time point. The involved actuators and sensors thereby interfere with each other. An asynchronous communication would help to avoid deadlocks.