Physikit: Data Engagement Through Physical Ambient Visualizations in the Home
Physikit: Data Engagement Through Physical
Ambient Visualizations in the Home
a) Community driven sensor kits  provide rich information about the environment and include data such as noise, air quality, temperature, or traffic.
b) While these sensor kits democratize urban sensing, data is often only available through websites or public datasheets with little support on how to use or interpret it.
c) To empower and provide users with better ways to interact with data, we propose a ‘human-data design’ approach that bridges the gap between non-expert users and their data.
d) Research has shown that physical and tangible interfaces can increase awareness and participation through their physical properties and affordances
e) Based on this work, we argue that providing physical, tangible and reconfigurable “physicalizations”  that match people’s own needs and interests, will encourage them to discover and understand the meaning of the data they collect and decide for themselves how to best use and share it.
f) we developed Physikit as a toolkit and technology probe  that makes users’ data visible and tangible through physical and embedded data visualizations called PhysiCubes
g) Physikit consists of
(i) a number of PhysiCubes that each provide one unique physical visualization such as movement, light, air or vibration, and
(ii) a web-based end-user configuration tool that allows users to quickly and easily connect data sources (Figure 1, left) to the PhysiCubes using a touch-enabled interface
h) Users can explore, interpret and engage with different kinds of data by creating simple rules for a variety of physical ambient visualizations.
i) A key research question this raises is: whether allowing users to program the mapping and relation between data and physical visualizations empowers them to explore, use and engage with data.
a) Understanding Data
The other households monitored the cubes and when the changes were frequent enough, they used the tablet to visit the website and look at the data represented by the cube. In these situations the cube itself did not provide enough capabilities to represent all the data changes, but was used as a catalyst that drew people in more to explore and understand the data.
b) Data To Cubes Mappings
The study demonstrates how participants created a diversity
of rules but also indicated what was of most interest to them.
c) Flow-Based Programming
i. The study suggests that people are comfortable with, and understand readily a guided form of end-user programming when configuring domestic-based IoT.
ii. The well-established pipe-based programming paradigm is an easy to use and understandable interface for configuration of IoT devices.
iii. While the back-end does support complex data aggregation, and combined outputs, the interface does not provide programming concepts and UI components to support this
iv. There is, thus, an important trade-off between configurability and ease of use that needs further research to understand the difference between guided and free-form data exploration, but also to explore programming of mappings through the cubes themselves.
d) Integration and Appropriation
i. All households creatively integrated the cubes into their
homes to find suitable use cases for the visualization in relation
to the available data.
ii. In general, our study indicates that using physical
tangible displays provides much scope for exploring data.
iii. The current study explored the use of Physikit for
a short period of time, but more studies are needed to investigate the long-term effects and sustainability of ambient physical visualizations for various data sets.
Intro: Our study demonstrates how 5 households used Physikit to explore a variety of data collected in their homes through configuring, appropriating and integrating the physical ambient visualizations into their everyday life. The results indicate that Physikit allows people to craft their own experiences, affordances and interpretation of data to help them build an engagement with the data set.
Physikit builds on four strands of related work:
(i) tangible user interfaces (TUI),
(ii) ambient information systems,
(iii) physical visualizations, and
(iv) end-user programming.
a) Tangible Interfaces
(1) - Tangibles have been used as remote control units for media [7, 12], in which interaction with the physical artifact is leveraged to control a remote installation.
A second class of TUIs such as AutoHan , Cognitive cubes  and AudioCubes  use tangible bits as programming or control input to construct objects on a computer device.
They use physical shape, form, and flexible connections between the cubes as buildings blocks to create a rudimentary vocabulary for designing new objects.
Because physical cubical building blocks are compelling embodiments of more complex abstractions, they have been frequently leveraged as a learning and exploration tool.
(2) “The Cubes”  uses a set of networked tangibles that can be connected for a game-based learning system.
(3) More recently, Chung et al.  introduced Cubement, a tool consisting of connected cubes used to create moving physical computing interfaces
(4) However, the input bandwidth provided by tangibles alone is limited compared to traditional or sensor input. Although their physicality helps in transforming data from the digital into the physical world, tangibles alone often cannot provide users with all the tools needed to effectively explore and use diverse data streams. However, by building on the principles and ideas of reconfigurable tangibles in data representation, we can revise physical tangible artifacts to primarily function as output devices for data.
b) Ambient Displays
(1) - Parallel to TUIs, ambient information systems were introduced as systems that visualize abstract interpretations of data in the environment or the user’s periphery of attention
Ambient systems “present information within a space through subtle changes in light, sound, and movement, which can be processed in the background of awareness”
(2) - Early ambient systems, such as ambientRoom , Audio Aura  and “The Information Percolator”  focused primarily on showing digital information through output modalities that integrate into the environment.
They leverage ambient light, auditory sound cues and output, such as water fountains, to visualize and provide peripheral awareness on peoples’ activities, information or social connections.
(3) - Cubble , for example, uses a cube that lights up, vibrates or heats up to allow people in a long-distance relationship to communicate. Similar moveable artifacts are the moving post-it notes by Probst et al.  that can be placed in the periphery and can be activated to draw people’s attention when needed.
Often, these ambient systems rely on ephemeral interfaces  that have a strong temporal focus and use tangible output that appeals to human senses, e.g., sound, air, light, or water.
(4) However, ambient displays are frequently used as passive portals into the digital space that do not encode complex data that can be tailored and appropriated by end-users as part of a data exploration.
c) Physical Visualizations
Such a data physicalization encodes data in its geometry or material properties
Many physical visualizations focus on a direct mapping between the data and representations.
A second class of physical visualizations are data sculptures
, which encode data using aesthetic features that push the
physicalization beyond a mere representation of the data, to
an artifact with sociocultural significance. An example of a
data sculpture is the Water Lamp , which encodes physical
bits as light-based water ripples
All of these examples show the potential for mapping everyday objects onto socially meaningful events
However, existing visualizations either focus on personal abstract representations that are not necessarily connected to the original data (data sculptures), or move the data sense making problem into the real world (physicalizations).
More research is needed to determine how physical visualizations can be used to
(i) enable users to explore and interrogate data themselves, and (ii) elicit interest to spawn actionable insights about data.
d) Pipe-Based End-User Configuration
Previous work has proposed ‘pipe-based’ end-user programming
to configure or program interfaces through a visual editor
that allows users to connect object by drawing pipes
2.Physikit decouples the input from the tangible cubes and adopts this pipe-based configuration concept to allow users to connect the sensor data to the cubes and configure the visualization from their mobile device.
The pipe-based approach was chosen because of the
clearly defined input and output space of the toolkit
Five households from London and South East UK participated
in a two-week field deployment.
Each participating household was given a set of four cubes
(Light, Buzz, Move and Air) and a Smart Citizen kit (SCK).
WiFi base station that was used to connect
the SCK and the PhysiCubes to the Physikit web platform.
i. The field study consisted of two phases. First, after an induction
to the study, signing an informed consent, and collecting
demographics, the households were interviewed about their
current knowledge and perspective on sensor data. Next the, households were given a SCK to deploy in their house
ii. Participants did not receive the PhysiCubes at this stage because we wanted them initially to get accustomed to the sensors in the kit.
iii. Second, after 4 to 5 days, the families were interviewed
to probe their insights on the sensor data of the SCK.
After this second interview, we demonstrated the Physikit
toolkit to the households until they were familiar with its
iv. Interactions on the application and all SCK data were logged. We collected qualitative data using
(i) experience sampling via diaries,
and (ii) interviews and contextual inquiries at the participants’
intro: The goal of the field study was to investigate
(i) which input-output connections participants would make,
(ii) how they leveraged the cubes to explore, use and understand the Smart Citizen kit (SCK) data, and (iii) how they would appropriate, embellish and craft experiences to integrate the cubes into their homes and everyday routines
i) Despite the increased availability of low cost sensors, most of the produced data is ‘black box’ in nature: users often do not know how to access or interpret data.
ii) This paper introduces Physikit, a system designed to allow users to explore and engage with environmental data through physical ambient visualizations.
iii) present a two-week field study which showed that participants got an increased sense of the meaning of data, embellished and appropriated the basic visualizations to make them blend into their homes, and used the visualizations as a probe for community engagement and social behavior.
a) Use Patterns
The data shows that the light cube was clearly the most popular visualization
However, overall the light cube did not substantially outweigh other cubes, as the use depended greatly on what sensor data people were interested in.However, overall the light cube did not substantially outweigh other cubes, as the use depended greatly on what sensor data people were interested in.
PhysiLight cube was most often connected to the light sensor and the humidity sensor and much less to other sensors.
For the PhysiAir cube, most connections were made to th humidity sensor and the NO2 sensor
The PhysiMove cube was most often connected to the temperature, and the noise sensors
The PhysiBuzz was mostly used with the noise and temperature data
In general, users were most interested in noise data , followed by humidity, temperature and light
Looking at the type of mapping (alert, continuous or relative), there are clear differences between the cubes.
Figure 8 shows usage patterns can be categorized into three main approaches:
Fixed Connection: one data rule was created and used
throughout the entire deployment (e.g., PhysiMove, h1).
Rapid Early Exploration: short early data rule changes
leading to a fixed and long-term data rule configuration
for the rest of the deployment (e.g., PhysiAir cube, h2).
Continuous Explorations: short and long iterative explorations
throughout the study. The data shows both homogenous
explorations using the same sensor (e.g.,
PhysiLight cube, h5) and heterogeneous explorations
switching between sensors (e.g., PhysiLight cube, h4).
b) User Experience, Appropriation, and Use
showed how Physikit made data more visible to users, resulting
in a change and broader interest in the different types of
Overall, the findings suggest that all households were engaged with the PhysiCubes and created a large number of rules to explore the data in a range of creative ways.
The understandability and memorability of the mappings between what was being sensed and what it meant was sometimes ambiguous requiring them to physically annotate and appropriate the cubes.
AIMS AND OBJECTIVES
a) Physical Visualization
Physikit provides physical and embedded ambient data visualizations
(PhysiCubes), which visualize one unique data
source through physical dynamic output
This notion of an atomic visualization ensures that the cube communicates the data source in an unambiguous output
format as set up by the user
Each PhysiCube visualization has an output range from 0 to the maximum value of the output visualization. Sensed input data is mapped to the output range of the visualization.
a. Independent of the physical visualization that is built into the
PhysiCube, each cube has the ability to visualize the data
through three different input-to-output mappings:
. Continuous: the sensed data is mapped linearly and continuously to the output of the visualization.
Relative: the output of the cube shows relative changes
in the data. Relative changes occur in both positive and
negative directions, signaling changing trends in data.
Alert: the configured output of the PhysiCube visualizes
an event when a threshold value that is set by the user is
reached by the data. v
Using these three mapping types as the vocabulary of all
PhysiCubes ensures an operation consistency across physical
b. Independent from the data or output visualization,
each cube will respond in the same way when new
data is visualized
c. Independent of mapping type, the visualization
will be run once when new data is visualized.
d. This consistency gives users a stable concept for the exploration of combinations of input and output to build a mental model of the relation between the data and the output of the PhysiCubes.
e. Physikit, thus, allows only for system- or synthetic interactions,
but not for physical interaction
a.The base design of all PhysiCubes is a cubical artifact that is equipped with brackets and hooks that can be used to attach other artifacts, materials or objects, or to attach the PhysiCube to the environment.
PhysiLight : visualizes data through a matrix of RGB LEDs.
PhysiBuzz: visualizes data through vibro-tactile feedback provided by six vibration motors.
PhysiMove: visualizes data through movement of a disk at the top plane of the cube. In its current form it is shaped like a star.
PhysiAir: visualizes data through airflows produced by a small and large fan.
Physikit in its current implementation provides four cubes: PhysiLight, PhysiBuzz, PhysiMove and PhysiAir
Each PhysiCube visualizes the data in a distinct way: through light, vibrations, movement, or air flow
All cubes can visualize data through a continuous mapping, visualize an alert whenever a configured threshold value is reached, or notify users whenever relative sensor data changes in positive or negative directions occur
Continuous and relative changes are visualized constantly, while alerts are only triggered one time when new data arrives.
b) Configurability and Guided Exploration
End-User Programming Interface
a. To enable users to create and visually explore the connections between input (Smart Citizen sensors) and output (the PhysiCubes), a web-based cross-platform end-user programming tool was developed
b. Through a set of steps, users create data rules that define how Physikit visualizes the input data on the PhysiCubes.
c. When a new connection
is made, the tool shows the users three input boxes that
ask the user for details to help them configure the rule
d. The first input screen provides the user with the option to select the type of mapping (continuous, relative or alert) between
input and output
e. After confirming the mapping
type, the user is prompted with a second screen allowing
them to configure the mapping. Only for the “alert” mapping
users can provide specific trigger conditions.
f. After selecting and configuring
the mapping, users decide in a final input screen
(Figure 4C) how the mapping is visualized on the output
g. This means they can use the PhysiCubes to visualize both their own and other peoples’ data to allow them to compare data, e.g., to check if their neighbors are noisier than they are.
To explore the relation between data and output visualizations, Physikit provides a user interface that allows users to configure the relation between data and cubes on two levels (see next)
Together, the connection and mapping form a data rule that describes how the cubes visualize the data input.
Using an end-user programming interface, users create data rules to help them explore the underlying data source
Although a full inputoutput space allows advanced mappings and relations between the Smart Citizen data and the PhysiCubes, the system supports constraints and guidance metrics to provide users with limitations and a path of least resistance in data explorations.
To support users in understanding how to create data rules for particular data sets, Physikit provides abstractions, thresholds and benchmark values.
The type of abstractions include:
Input and Output limitations: which mediate and limit input and output values that lie beyond the range of the PhysiCube output modality and input data set.
Pipeline limitations: that constrain the amount of connections made between input data and physical visualizations to ensure unambiguous visualization of data.
Abstract representations: rather than presenting the user with raw data, the toolkit provides abstract representations of value ranges in the form of symbols, concepts or other understandable representations.
Input-Output Connection: users can create pipeline connections between data input and visualizations on the PhysiCubes to define a relation between in- and output.
Input-Output Mapping: once the user defines a connection
between an input data set and visualization, they can
determine the mapping (continuous, relative or alert)
and behaviors (type of output) of the cubes.
i. The cubes are the same handheld-size (11x11x11cm) and
come in different colors to make them distinguishable.
ii. Physikit uses a Node.js website that provides user- and data rule management through a web-socket connection
iii. It connects to the Smart Citizen API and processes and cleans sensor data before pushing it to a rule engine that calculates the input-output mapping that controls the individual cubes via the web.
The aim of our research is to explore how configurable physical
ambient visualizations can be mapped onto sensed data
to become ambient data objects.
The Physikit toolkit is designed to work with any data, including
environmental, personal, or health data. For the
study reported here, we chose to investigate how households
would explore and understand environmental data collected
in their homes using Smart Citizen
One of the problems with the way Smart Citizen is currently
set up is that many people find it difficult to understand the
Users also reported that the kit itself became invisible
after a while, resulting in people losing interest in the
data and the kit.
This provided us with an opportunity to explore
how to make the data more meaningful. By providing
another layer of physical ambient visualizations – that users
themselves program to map onto the visualized data
The central goal of Physikit is to represent data via physical
ambient artifacts that can be programmed and configured by
a non-expert user.
Physikit was designed as a new kind of interface for the general public to explore data through reconfigurable and appropriable physical ambient visualizations that represent data through movement, vibrations, air and light.
Physikit has shown how it is possible to democratize data to the general public in ways that are meaningful, creative, and aesthetic, while opening the door for end-user programming to be repurposed in the realm of IoT.