Venous Materials: Towards Interactive Fluidic Mechanisms

HCI researchers have investigated how to shift interfaces away from screens and into physical objects and materials [23, 25, 69, 80]. While researchers commonly develop such interactive materials with computers combined with sensors and actuators [10, 23, 24, 32, 35], such systems rely on rigid and bulky hardware as well as real-time control software.

To tackle the complexity of these systems, HCI researchers have sought to develop non-rigid interactive material with means of computational fabrication, soft robotics, and material sciences [4, 5, 54]. Pneumatics have been widely explored for shape and stiffness changing interfaces [44, 45, 84]. In such research, the transformation pattern was embedded in the soft inflatable composite through simulation and fabrication methods. Generally, pneumatics still rely on rigid and bulky air pumps and micro-controllers, so researchers have explored embedding the interactive capabilities by encoding responsive behavior into material structures themselves. For example, by using hygromorphic or thermoplastic materials, researchers have proposed a method to design responsive shape changing behavior reacting to everyday interactions such as human-body sweat, heat of lamps, or even boiling water [11, 73, 74, 78, 85]. Another approach of material design with embedded interactivity is by designing the mechanical deformation properties that responds to external deformation and force. Metamaterial Mechanisms and KinetiX took this approach with architected hinge-mechanism to create intended deformation [20, 43]. Ion et al. extended their work to encode computational logic, and texture change to the fabricated material which expanded its interactivity [21, 22]. HCI researchers have shown to program not only the mechanical deformation by design of internal structures, but also other interactive properties, such as programmable buoyancy, center of mass, and sound of wind instruments [2, 52, 68, 77].

The key idea for this branch of research is the introduction of fabrication methods and simulation models to create a responsive material that is encoded in the structure and property of material itself. This research space is defined and explored along multiple research concepts/paradigms, including Metamaterials [20, 48], 4D Printing [64] and Smart Materials [62, 70, 75, 83]. As for such overlapped research realms, our work is intended to bring a novel approach of fabricating fluidic path within soft materials that is designed to be responsive to tangible and deformation input. This paper specifically seeks to explore mechanical deformation as input and color change as output as a first step towards this goal.

Color changing material have also been explored in HCI to enrich the interactivity of physical matter using triggers such as temperature [31, 40, 56, 72] and UV light [16, 29, 53, 58]. In the materials science field, color changing materials responding to deformation has been explored as well [13, 59]. Our approach of using fluidic flow has a variety of design parameters for display patterns, mixing of colors and deformation inputs.

Recent HCI and Kinetic Art projects utilize fluids for interaction design and display [50, 15, 37, 41, 60, 65, 67, 71]. Several projects explore computational control of the liquid flow in tubes and other fluidic paths to display information [51, 19, 38], while we aim at controlling the fluid flow within a passive material reacting to deformation without any pumps.

In recent research in materials science, mechanical engineering, and computational fluidic systems, Microfluidics has been a topic that is widely explored [82]. For example, in Robotics, Microfluidics was applied to change the color of a soft robot for camouflage [39]. Microfludics with a chemically responsive material was used to develop robots with an entirely flexible body without any electric components [79]. SkinFlow utilizes transmission of liquids in a series of chambers to detect object deformation but with a camera [63]. Garrad et al. proposed a method to encode fluidic logic into a soft robotic system using conductive fluids inspired by biological veins [14]. One of their prototype integrated a transformation triggered by touch input with fluidic channel via electric actuation, which demands battery components.

There have been efforts to develop methods for researchers and designers to construct and customize their own microfluidics structures [8, 46, 66]. In our paper, we share our design and fabrication method utilizing custom developed simulation software as well as laser-cutter-based fabrication technique to design variety of responsive fluidic structures. Comparing to other simulation software of microfluidics [28], our software is intended to simulate the fluidic movement in reaction to deformable inputs for interaction design. While our design and fabrication was developed through a hands-on iterative process, we aim to not only convey the vision and concept of Venous Materials, but also to inspire designers and researchers to develop responsive fluidic materials with a relatively accessible method.

We first define the basic architecture of Venous Materials as shown in top of the Figure 2. Venous Materials is chiefly composed by two types of materials; Substrate Material, which is the external deformable substrate that has enclosed fluidic channels (here we use PDMS), and Liquid Material, which is the internally contained fluid that dynamically flows when deformation is applied (here we use water based ink). For the internal structure of Substrate Material where Liquid Material is contained, there are three major components. First, the Liquid Repository is designed to store the liquid in one area. In order for the fluid to flow out of the repository, the deformation and applied pressure should be focused mainly in the Liquid Repository. Second, the Fluid Channel is designed to connect to the Liquid Repository in an arbitrary path and geometry. The design of this path defines the display pattern of the material. Last, Air Repository is designed to be placed on the end of the fluid channel to control the air pressure of the internal structure. This enclosed Air Repository, is the key parameter that enables the fluid to flow back to the liquid repository after each deformation. The Air Repository can also be open to atmosphere, instead of enclosed in the Substrate Material which will make the flow irreversible after the deformation.

Size and dimensional ratio of the Liquid Repository, Fluid Channel, and Air Repository defines the sensitivity and scale of how the liquid flow in reaction to the magnitude of deformation. In our paper, for example, Liquid Repository is in milli-scale to contain a volume of Liquid Material, and Fluid Channel is in micro-scale to display dynamic movements of liquid flow. The way such a dimensional ratio affects the interactive properties motivated us to develop a design tool which can simulate the fluidic flow by taking into account the ratio of fluidic mechanism.

With this basic architecture, there are a variety of properties that can be customized and configured for each Venous Materials devices to create different types of interactive designs. For example, Substrate Material can take arbitrary 3D shapes to provide richer affordance design, or ferromagnetic liquid can be used as Liquid Material for a non-contact actuation. While, in this paper, we focus on the basic internal channel design that affects the fluid motion within a flat substrate material with colored liquid for visual output, we believe this basic architecture leads to much wider design space. The properties explored in this paper and the ones for future opportunity are represented in the basic architecture diagram in Figure 2, as the properties explored in this paper are highlighted with yellow.

Based on this core architecture, the rest of the design space presents the extended design primitives in four categories; Flow and Display, Deformation Inputs, Composition and Functional Utilities. This design space is the core of this paper that bridges across following sections of the paper; Flow and Display as well as Deformation Input will be referred in our design pipeline, especially for simulation software design, and Functional Utilities will be demonstrated through multiple applications.

One of the most important elements in Venous Materials is the capability of displaying dynamic information with visible flow. We define this display capability by introducing Flow Primitives, which demonstrates how a primitive channel can create different types of flow, Geometry Patterns, which describes how the Flow Primitives can be composed to create different geometric pattern, and Multi-Color Primitives, which utilizes multi-color in the fluidic channel.

Geometry Patterns. Based on the Flow Primitives, we designed multiple Geometry Patterns that offers different visual display outputs in the design, as pictured in Figure 3. As these patterns were designed to be utilized as a pallet (or swatches) by users to create complex fluidic geometry, the presented patterns are examples of many other possible ones. The different patterns are described in few major categories below.

Figure 3: Sample Geometric Patterns (FRACTAL, BRANCHES, VEINS, MESH and FREE LINE).

The simplest category of Geometry Pattern is the continuous single line (e.g. LINE, SPIRAL, FRACTAL, and WAVE). This type of geometry simply maps the magnitude of deformation input to how far fluid travels in the single stream of flow. There is a category of patterns that utilizes the flow primitives, Split and Merge, to translate a deformation input to a dispersion display effect (e.g. RADIAL, BRANCH, and VEIN). These patterns can be utilized to create a display effect that can spread across a large area. Finally, there is a group of patterns that utilizes Intersect flow primitives. For example, MESH makes many intersections within a grid geometry, while FREE LINE makes intersections within a single continuous line. With this pattern, the Intersect primitive enables irreversible effect as shown in Figure 4 a.

Figure 4: Two types of irreversible samples with a: MESH (Once the flow pass through the mesh, some of the fluid remains trapped in the channel, in three levels.), b: Color-Mixing (the mixed color is irreversible).

Multi-Color Primitives. Another important aspect for display capability is the color. Fluid with different colors can be used for Mixing to make color changes in response to deformation (Figure 4 b). This primitive is irreversible, which can be applied as a material with a mechanical memory. For reversible Multi-Color Primitives, fluidic channels with different colors can be composed in different ways including a Gradient by configuring the channels in order of gradient colors (Figure 5 a), Overlayed to each other (Figure 5 b) or Interweaved with multi-layer composition.

Figure 5: Multi-Color Primitives. a: Gradation, and b: Overlay.

Various ways of deformations can be applied as input to the fluidic mechanisms (Figure 6). This element is a key to trigger the fluid to flow within the channel. We define four major types of Deformation Inputs in Venous Materials; Pressure, Bend, Stretch (i.e. tension) and Twist (i.e. torsion). For each input, the magnitude of deformation affects the physical motion of liquid.

Figure 6: Deformation Inputs, Pressure for a: Point, b: Area, c: Slider, and d: Bending, e: Twisting, f: Stretching.

Pressure input simply represents an input made by pressing the material. Pressure can be applied in different types of spatial modalities, including Point (i.e. pressing with a finger tip), or Area (i.e. applying force with a palm or foot). The pressure effect can also be shifted across a continuous surface which can be described as Slide. For the Pressure input, the magnitudes as well as location of the pressure affects the fluid motion.

Input actions that are afforded by Venous Materials’ deformable sheet form factor can also be the trigger for fluid flow. Bend is one such example. The bending curvature and angle can be mapped to the magnitude of flow within the channel. Twisting and Stretch can also trigger the flow when applied as deformable inputs.

As Venous Materials are composed of layers of flat sheets, different types of Composition can extend its simple structure and expand display and interaction capabilities.

The first type of composition is the use of multiple layers. While Single Layer Venous Materials is relatively easy to fabricate, it has certain limitations (e.g. the liquid cannot be overlayed without mixing). With Isolated Multi-Layer composition, Venous Materials can have Overlayed multi-colors, or with Connected Multi-Layer, the liquid can move across multiple layers to create complex pattern like the Interweaved multi-colors.

The second type of composition is the I/O Spatial Composition. The location of deformation input relative to that of the fluid display can be composed in either Co-located on the same location, or Transmitted apart from input. The composition of Co-located can be useful for users to intuitively understand the relationship of I/O, while Transmitted can be useful for users to see the display that is not occluded by the deformation source (e.g. pressure under a shoe).

We define how Venous Materials can be utilized for interaction design as Functional Utilities. This section describes how the responsive flow and display capabilities of Venous Materials can be interpreted and used for functional purposes in HCI and interaction design.

We implemented a specialized design tool which enables users to model, simulate, and adjust the venous structures. Users can simulate the flow response and fluid distribution under different deformation inputs and export the files for the laser cutting / engraving process.

The way Venous Materials respond to a deformation input is affected by different parameters of the materials and geometries. The main motivation of the GUI design tool is to provide users with a method for real-time simulation to support the iterative design of functional fluidic mechanisms.

Researchers in HCI have previously shown the benefits of design of specialized computational editors for simulation and design of programmable interactive materials [20, 21, 42, 44, 74, 85]. Our design tool builds upon this HCI stream, however, it is specifically designed for fluidic structures and flow. While computational simulations are increasing in popularity in the microfluidics field [28, 76], our design tool specifically focuses on the simulation of fluidic flow in reaction to deformable input for interaction design purposes.

Figure 9: Design and simulation workflow of the prototype on Figure 8.

Our design tool is implemented in Rhinoceros 6 and Grasshopper environment. In this section, we give an instruction through an example user workflow. Figure 9 illustrates the step-by-step design and simulation workflow for our example prototype; a double layered mechanism that displays two different colors in response to bending. Each step in Figure 9 shows only the relevant section of our GUI, instead of showing the full GUI window.

(a) Plan and Design: A user starts the design by choosing a linear primitive geometry pattern that the user would like to generate from the swatch of geometry patterns. Once the desired geometry pattern is chosen, the user can configure the geometry in two simple steps. First, the user draws and defines the area in which the geometry will be generated. This step allows them to easily design and generate patterns across large areas. Second, the user defines the flow direction by setting a starting-point of the flow (and in some cases an end-point as well). Once the two steps are done, the chosen primitive is automatically generated and displayed on the design view. The design tool is also capable of generating different types of geometries in a single model by repeating the steps above. Additionally, the software offers a free line drawing function where users can compose arbitrary linear shape of fluidic channel with a cursor. This function also lets users connect multiple geometric patterns.

(b) Adjust Parameters: Next, the user can adjust a set of parameters to define the detail parameters of the geometry of fluidic path. There are two major categories of parameters the user can adjust. One is the parameter of the primitives in the basic architecture including Liquid Repository, Fluid Channel and Air Repository. Properties including dimensions (e.g. size, width, and depth) for each primitives can be adjusted. The other category of parameters is for each geometry pattern, which are variables defining the global geometric configuration rather than primitive property of channels. The types of parameter depends on each geometry pattern, but they include density of patterns, orientation of the channels / patterns, etc. In this example, the design and parameter adjustment steps are repeated to generate a second layer.

(c) Set Deformation: When the design of the fluidic structure is complete, we set a virtual deformation input to the model of the fluidic channels. For folding deformation, as shown in this workflow, the user defines a folding line, folding curvature and folding angle. For pressure, the user defines a pressing point, a radial range of pressure, and a force.

(d) Simulate and Display: To simulate, the user turns on a simulation mode to display a Pressure Map and a Flow Map as a result. The Pressure Map view allows users to see a color-gradient map of the gradient of pressure intensity that results from the deformation input. In the Flow Map, users can view the simulation result of fluid distribution through the channels according to the deformation input parameters. This also allows users to view the change of overlapped colors in multi-layer composition. During the simulation mode, users can still modify any parameters from (a)-(c) steps. Users can view real time updates of the simulation and make iterations on their design. Finally, the design tool generates multiple DXF files depending on the fluidic channels and repositories with different depth. This is useful for the laser engraving step where each depth should be processed separately.

Multi-Layers: Our design tool supports multi-layer structure as well. The user can design and adjust primitives in different layers in the design step and also manually define connection between multiple layers. Simulation and actuation can also be applied for multiple layers together.

Fluid Volume Verification: Our design tool supports users to verify the volume of the fluid repository that can contain a sufficient amount of fluid for intended flow despersion. The user may take this parameter in consideration during the step (b), when simulation result of Flow Map view lacks the distribution of liquid flow.

Sensitivity Control: An additional parameter in the design tool is the control of sensitivity, which we define as the flow-response-ratio according to deformation input. The sensitivity is defined by the ratio between the volume of the channels to the volume of the air repositories. To realize this function, when the sensitivity value is tuned by the user, the system adjusts the volume of the air repositories accordingly.

In this section we explain the fabrication process of the Venous Material structures. The key points for our choice of fabrication method and materials focused on a quick, low-cost, accessible, and scalable process. The material we used as the substrate is PDMS (Sylgard 184), which is widely used in the microfluidics field and allows us to follow existing published protocols on mixing, casting and molding [1].

A basic structure of Venous Materials is composed with three layers of PDMS; a bottom layer with the fluidic channels and repositories, a middle layer that has a hole for fluid injection, and a top layer that is a plain sheet for sealing. For multi-layer composition, multiple sets of bottom and middle layers were stacked with single top layer on the top. In our prototypes, we used 0.9 or 1.5 mm thick PDMS for bottom layer, and 0.5 or 0.9 mm thickness one for middle and top layer.

The fabrication process starts with creating the PDMS sheets by mixing, degassing and casting into prepared molds. We then cure the PDMS by heating to 27 °C for 40 minutes, or leaving at room temperature for 24 hours.

We use a CO2 laser cutter and engraver (GCC Laserpro Spirit) for fabricating internal structure. This is the process to engrave fluid channel and repositories on the bottom layer. Based on previous characterization of laser cut PDMS [18, 36], we created our own library of channel and repository samples by changing engraving parameters, that affect the surface texture and fluid flow. Figure 10 shows a set of parameters that we optimized for our process. Other fixed parameters are: manual color fill mode, 1000 DPI, 200 PPI, and raster activated.

Figure 10: Laser engraving parameters library.

After laser engraving, the engraved PDMS has to be cleaned with isopropyl alcohol (IPA) to remove grease, and tape to remove dust. We manually punch a hole to the middle PDMS layer based on the position of the fluid repository accordingly. Then, we manually align and layer the bottom and middle layer. We used two different methods for layering and bonding the layers in this process. The first is simply utilizing its self-adhesive properties, useful for rapid prototyping and re-injection of the fluids. The second method is using plasma bonding for higher adhesion between the layers. This is more efficient for condensed patterns and larger deformation input forces. After the layering we manually inject the fluid (Ecoline water-based inks) through the holes on the middle layer. Lastly, the whole composition is completed with seal using top layer of PDMS.

Figure 11: Simulation workflow. (a) Division into cells. (b) Calculation of resistant pressure map. (c) Calculation of actuated pressure map.(d) Calculation of flow map.

This section gives a brief description of the workflow of our simulation method that was part of our design and simulation tool (See Figure 11).

Division into Cells: We first divide the channels into cells and use a data tree structure to store them. For each cell, we use a coordinate (m, n) to describe its location, where m is the index for the branch where the cell is located in the tree structure, and n is the sequence index for the cell inside the branch (Figure 11 a).

Calculation of Resistant Pressure Map: Next, we define a Resistant Pressure Map $P_{(m,n)}^{r}$ to evaluate the resistance of the venous structure to fluid movement. This resistance comes from an increase in the internal pressure when the fluid flows further into the channel, and compressing the air. We use Boyle‘s Law to calculate the internal pressure $P_{(m,n)}^{r}$ for each cell (m, n) at the moment when the fluid reach that cell, which represents the magnitude of the internal pressure that is required in order for the fluid to travel to the cell (Figure 11 b).

Calculation of Actuated Pressure Map: Then, we define an Actuated Pressure Map $P_{(m,n)}^{a}$ to evaluate the increment of internal pressure caused by the external actuation. The key point in this step is to calculate the incremental internal pressure of the fluid repositories $P_{fluid\ repo}^a$. We make this assertion for two reasons: 1) the fluid repositories are the source of a flow map, so in most cases the actuation is applied close to the fluid repositories; 2) we assume that we could equalize the fluid pressure in the channels to the fluid pressure in fluid repositories, as long as they are connected together, so that we can get the whole actuated pressure map $P_{(m,n)}^{a}$ after calculating $P_{fluid\ repo}^a$.

To get $P_{fluid\ repo}^a$, we use computational models to characterize the relation of the incremental internal pressure P^a to the actuation parameters, which we selected as pressing pressure P_press and folding curvature $R_{fold}^{-1}$. In the computational models, we first infer the proportional relation of P^a to P_press and $R_{fold}^{-1}$ with fundamental physics analysis, and next introduce a Gauss factor to describe the decay of the effect of actuation caused by the plasticity of PDMS. The integrated computational models are shown as Equation (1)(2), where k_p and k_f are two proportional coefficients, d is the distance off the central actuation position and σ is a decay coefficient.

Figure 12: Experiments to test the resistance property of the venous structure on the fluid movement: (a) Experiment results. (b) Simulation results.

Figure 13: Experiments setups. Top:experiment of folding actuation. (a) Experiment set up. (b) Side view of experiment process. (c) Top view of experiment process, and data collection of folding curvature $R_{fold}^{-1}$ and fluid travel distance l. Bottom: experiment of pressing actuation. (d) Experiment set up. (e) Experiment process, and data collection of pressure P_press and fluid travel distance l.

Figure 14: (a) Fitting results of P^a − P_press curve from one device in pressing actuation experiment. (b) Fitting results of $P^a-R_{fold}^{-1}$ curve from one device in folding actuation experiment. (c) Estimate results of the undetermined coefficients in the computational models (Equation 1,2).

Calculation of Flow Map: Once we have both the resistant pressure map $P_{\left(m,n\right)}^r$ and actuated pressure map $P_{\left(m,n\right)}^a$, we use an iteration algorithm to calculate the flow map F_{(m, n)} (Figure 11 d). For each cell (m, n), the solving process is shown as Equation (3) where F_{(m, n)} is a Boolean value representing whether the cell (m, n) is filled by fluid, precursor of (m, n) is the adjacent precursory cell of (m, n) in the data tree.

We designed a few experiments to test the models we used in simulation, and to estimate the undetermined parameters. The experiments were conducted in the following stages:

• Experiments on resistant pressure map: test the resistance property of the venous structure on the fluid movement.
  
• Experiments on actuated pressure map: test the computational models (Equation 1,2) and estimate the undetermined parameters.
  

Experiments on Resistant Pressure Map. We conducted an experiment to test the resistance property of the venous structure on the fluid movement. We used an asymmetric branch structure as our experimental device and applied a pressure gradually to the central fluid repository. We conducted this process both in our design tool and in an actual experimental environment. We then compared the simulation result with the experimental result. Figure 12 shows the comparison, where we find our simulated changing process is in close agreement with the actual changing process. This result shows a high coincidence of the resistance property between the simulated and actual performance, which proves the validity our resistant pressure map model.

Figure 15: Applications of Venous Materials includes visualisation of (a, b) pressure distribution for shoe-soles, (c) bending angle of finger joints, (d, e) pressure applied to piano keys and brushes, (f) internal object condition of a package.

Venous Materials can dynamically visualize people's bodily actions, which can be used for embodied expression, learning, and augmentation. For example, the shoe-sole prototypes demonstrates the utility of embodied learning with body-worn Venous Materials (see Figure 15 a, b). This prototype is designed to visualize pressure applied on the shoe sole, while the fluidic display is located on the top of the shoe. It has the advantages being flexible and attachable, which makes it easy for users to attach or embed to their shoes. By visualizing the invisible pressure to users, this can be used to raise awareness of their body balance and motion for rehabilitation and dance lessons. Future challenges include improving visibility by scaling the display area.

As shown in Figure 15 c, not only pressure force but also bending motion of the body can be an input. As bending joints of bodies is crucially important for a variety of rehabilitation and training tasks, Venous Materials can be applied in such scenarios as passive responsive materials.

The Venous Materials can also be attached to physical objects and tools to display bodily interaction to them. The two following applications intend to give colorful expression to artistic motion such as playing the piano and drawing. While both drawing and playing music are artistic fields, learning process often requires an analytic observation on professionals during their performance. For augmenting piano keys (Figure 15 d) and a paint brush (Figure 15 e), by display of pressure and intensity can visualize additional layer of information to the student that observe the master. Furthermore, in piano, intensity and pressure has linear relations with the playing speed. We envision this tool being used earlier on in the learning process when users slowly press the keys in order to learn the appropriate striking force. Similar to Soter's approach to capture the fluid flow with camera system [63], the change of color on Venous Materials can be captured with a camera to digitally track a user's kinetic action for annotating, and archiving (Figure 15 d shows our prototype of such computer vision app).

There has been an industrial demand in tracking / indicating package status with passive devices [61]. Using the Venous Materials’ irreversible property and capability of pressure memory, users could track customized information such as the amount and specific location of force or impact applied to the package. This way, an object's status can be conveyed on the outside of its package so that deliverers or sales store representatives can be aware of the content's condition without opening the packaging (Figure 15 f). Real-time representation of internal weight distribution of package could also be useful to ensure the status of fragile products without opening it.

Figure 16: Application of Dynamic Information Representation for (a) Chinese Character Writing, (b) Line Chart.

As the value of dynamic data representation and physicalization is being considered in a recent paper [26], Venous Materials can be used to present dynamic data by encoding information in the material itself. Figure 16 a shows how the sequential stroke order for the Chinese character for ‘water’ can be dynamically visualized. Users can gradually apply force to fluidic repository to control and view the dynamic data of writing the character.

Figure 16 b presents an interactive line chart, that are activated and animated with pressure input by the user. Here, the sequential flow is representing time and location that can dynamically change. Users can also choose to display multiple curves at a time and compare the progression of the data and its intersection with other curves. Analog fluidic mechanism utilize the advantages of battery-free, self-contained, flexible, and thin nature of the system that allows it to be embedded into a text book, pop up book or children's book.

While the fabrication approach introduced in our paper is accessible with off-the-shelf equipment, one limitation is the fragility of the material. Porosity of PDMS is also an important challenge, as liquid tends to evaporate after a long period of usage. We are looking for more robust solutions to enable long-term usage, as well as greater precision, and accuracy for Venous Mechanisms. While our current fabrication pipeline relies on manual tasks (e.g. assembly, cleaning, and injection), there is future work at automation and mass production possibilities. Soft 3D printing is a promising direction for designing complex 3D volumetric fluidic structures without manual assembly, as they are becoming more accessible and precise [12]. Microfluidics’ mass production methods such as roll to roll microfluidics [17] is also an exciting direction.

More research needs to be done in order to realize large scale fluidic mechanisms. Future challenges include addressing the accumulating forces of the material on itself and high pressure in the channels. In terms of applications, scaling the display area will include improving visibility of the mechanism and open a space for larger applications including larger wearables (e.g. dance wear, or fashion) and interior products (e.g. carpet or wallpaper).

While our design tool allows relatively accurate simulation of fluidic mechanisms, there is a room for future improvements. Factors such as the surface condition of channels, local deformation of the substrate and the thickness of the device could be taken into consideration for more accurate and precise design. Similarly, connected multi-layer composition and intersects can currently be designed but not simulated. At this time, the design tool can only be used for pressure and bending input. Other types of deformation inputs can be addressed in future simulation tools and algorithms.

The future of fluidic mechanisms holds a rich interaction space for tangible and embodied interfaces with a device composed only of passive materials. As we discussed about the basic architecture and potential design properties of Venous Materials in Figure 2, there are many interaction design opportunities that may be extended with future design properties. Experimenting with different types of Liquid Material properties is a promising direction, as, in HCI, many research has been conducted to utilize material that reacts to heat [56], pH [30], electric field (using liquid metal / electrorheological fluids) [27, 37, 60, 65], magnetic field (magnetorheological fluids) [27, 57, 71, 81], or UV light [53]. We believe that by combining these special liquid materials with our fluid mechanism design, new interaction opportunities will open up not only for color changing displays, but also for dynamic behaviors in shape, stiffness, temperature and other interactive properties in response to tangible deformation inputs.

In addition to extending interactive properties with special material properties, deeper exploration of the computational logic enabled by fluidic structures is another exciting direction for Venous Materials. Microfluidics research holds an in-depth knowledge of complex integrated logic design [82]. By utilizing such knowledge, it could be extended to incorporate complex logic that would open up the interactivity of the material (e.g. counting the number of deformations, triggering different output patterns in response to the type of deformation inputs [IF THEN states], etc.). We envision a complex network of computational logic to also be embedded in the fluidic structures together with input and output functionalities. We believe that such potential capability of Venous Materials that fully couples I/O and computation enabled with composition of flexible and fluidic material would open up a new design paradigm for the domain of Interactive Matter and User Interfaces.