Soft Actuation for Home and Office

context-aware, non-intrusive hinting at simple yet worthwhile operations
on nearby objects

Proactive vs soft actuation.png

The prevalent vision of pervasive computing is that applications should operate proactively, i.e., (a) sense, (b) infer higher-level context, (c) decide how to affect the environment, and (d) actuate accordingly. The last step of this sense-and-react chain, actuation, is executed by the application itself. It is usually an operation on some object, e.g., switching on a light. This is exemplified in Fig. 1a, for a simple wintertime application that protects the user from catching a cold, by closing the window when the inside temperature drops below a threshold.

While such a proactive approach may appear to offer most added value (as the user does not have to do anything), it has serious pitfalls, indentified in the literature. There may be a number of reasons why proactive actuation may not be well received by the user. First, there may be “objective” errors in context sensing and inferencing, leading to obviously unreasonable actions. Second, the user may occasionally desire something opposite to what the system does. Third, the action can be unexpected and distractive. Finally, the user may feel a (possibly vague) sense of lack of control.

We explore an alternative to proactive actuation. We focus on context-based, application-generated hints that suggest the user to do specific actuating actions, which consist in reaching to a nearby object and performing a simple manual operation on it. Accordingly, a hint specifies both the object to be acted upon and the operation to be performed (e.g., “close the window”). We call the delivery of hints soft actuation, since the purpose of a hint is to trigger actuation, but the decision as well as the execution (if any) is left to the human.

Allowing the user to decide makes soft actuation unreliable. For any given hint, the application does not know whether the hard actuation will occur, and it may not have a direct way to know whether it has occurred. Clearly, the functionality of soft-actuating applications should not be life-critical.

We focus on the regular user in the home or office, as opposed to the professionally trained user in a specialized environment (e.g., the pilot in the cockpit). We pay special attention to the requirement that soft actuation should be non-intrusive. We cover the interaction technique in its entirety, i.e., the whole involvement of the human in the loop, including the decision to actuate and the actuating action, not just the delivery of hints.

We make the following contributions. First, we systematically describe the concept of soft actuation for home and office; while the high-level idea has been put forward quite some time ago, a systematic treatment is missing (to the best of our knowledge). Second, we position soft actuation with respect to related research areas. Finally, we suggest that soft actuation becomes the subject of a systematic study, identify selected research challenges to be addressed, and offer a few preliminary ideas as to potential solutions.



See our papers:

  • Domaszewicz, J.; Lalis, S.
    Soft Actuation for Home and Office
    Intelligent Environments (IE), 2013 9th International Conference on, pp.188-195, 16-19 July 2013, Athens, Greece, doi:10.1109/IE.2013.32
    pdf(paper), pdf(presentation), IEEE Xplore


  • Domaszewicz, J.; Lalis S.; Koutsoubelias M.; Tajmajer T.; Pruszkowski A.; Grigoropoulos N.
    Projekt SACCOM: walidacja "miękkiego sterowania" w środowisku biurowym
    (The SACCOM Project: validating soft actuation in an office environment, in Polish)
    XXIX Krajowe Sympozjum Telekomunikacji i Teleinformatyki KSTiT 2013, 4-6 September 2013, Gdańsk, Poland, and Przegląd Telekomunikacyjny i Wiadomości Telekomunikacyjne, pp.990-996 , nr 8-9/2013
    pdf(paper), pdf(presentation), PTiWTEL