\subsubsection{Discussion and conclusion}
-put together existing building blocks to design a whole system
-
-interdisciplinary skills and knowledge
-
-low vibration amplitude => clamping arrea => localized haptic feedback
-
+This work required interdisciplinary skills and knowledge to put together existing building blocks to design a whole system.
+This is what I consider a strength of Human-Cumputer Interaction research, at least the way I do it.
+We have skills in many research domains, which allows us to design, implement and evaluate interactive systems.
+When more expertise is required, we can efficiently collaborate with experts in other domains.
+In this project we collaborated with experts in material science who designed the tactile actuators modeled and simulated their vibrations in a theoretical environment.
+
+%Our colleagues who designed the tactile actuators modeled and simulated their vibrations in a theoretical environment.
+In these simulations, there was no finger, and the clamping of the surface was perfect, without tension.
+In practice, the finger dampens the vibration and it is unsure other vibrations disrupt the perceptions of the vibrations produced by the device.
+In particular the main scenario of the project\footnote{\href{https://cordis.europa.eu/project/id/645145}{H2020 Happiness}, grant agreement \#645145} was a car dashboard.
+Cars produce vibrations of various frequencies and amplitudes because of the engine and the road.
+Other partners of the project actually evaluated the perception of this tactile feedback in a real-case scenario~\cite{ng17}.
+On our side, we noticed the clamping of research prototypes is not trivial.
+It is difficult to glue the flexible substrate on a rigid surface with a clamping area so that the substrate remains perfectly flat on the clamping area.
+It creates a bi-stable condition on a surfaces that produces an undesirable click sensation when touching it with a finger.
+Therefore we must apply tension to hold the flexible substrate flat on the rigid surface.
+It is well known that adjusting the tension of a drum head on a drum shell changes its resonnant frequency.
+This is precisely the way we tune a drum.
+The same principle applies to our setup.
+However the vibration was still perceptible, with no bi-stable condition.
+
+A difficulty of this work was the long manufactoring process of actuators.
+It required weeks of planning, and the actuators were printed in a white clean room.
+This long process limited the number of iterations we could perform for designing the actuator layout and properties such as thickness, shapes and sizes.
+Therefore at each iteration we printed several configurations, then tested them to select the most appropriate one.
+However it limited the type of user studies we could perform.
+
+One of the major differences between this vibrotactile technology and other piezo-based actuators is that their thickness is very low.
+It is an advantage because it uses less materials.
+It is also a drawback because the amplitude of vibration is much lower.
+Therefore the vibration hardly transmits to thick surfaces like a 1mm thick plastic dashboard.
+The alternative is placing the actuators on holes that define a clamping area so that the substrate resonnates like a drum shell.
+This solution brings an interesting property that other technologies do not have.
+The vibration is localized to the clamping area, therefore it is localized.
+Hence with a dashboard with several buttons and sliders, it is possible to vibrate buttons individually.
+
+
+%slow iterations because long manufactoring process of actuators
+
+%low vibration amplitude => clamping arrea => localized haptic feedback
\subsection{Actuated computer peripherals}
projects / hdr.git / commitdiff