The Bicycle Barometer: Design and Evaluation of  Cyclist-Specific Interaction for a Public Display

The Bicycle Barometer: Design and Evaluation of Cyclist-Specific Interaction for a Public Display


– [Aaron] …the session, so thank you for coming back for after lunch. I’m Aaron Quigley from the
University of Saint Andrews. I’m your session chair. I’m a comedian. We have the first talk
on the bicycle barometer. This is work from Belgium, but
it’s going to be presented by Sandy on behalf of her
coauthors Andrew and Karen. Thank you very much Sandy. – Okay, thank you Aaron. So yeah, as Aaron said,
my name is Sandy Claes. I come from the research
and design group of the Department of Architecture
of the KU Leuven. And I will present this
work on a public interface that is specifically
designed for cyclists. So it’s design consists of a
pressure-sensitive floor mat, a hand rail that provides
cyclists with support, and also has a button mounted on, and a low-resolution LED display. So although we are in the session of large display interactions,
mine is actually very small. It’s a low-resolution screen, but the interaction is quite large, so maybe that compensates a bit. The outcome of this setup is
the result of an extensive user-centered design process, and we evaluated the outcome
in the wild for five days. And in the paper, we describe also eight design recommendations. So why this focus on cyclists? Well, cycling is being
promoted as an alternative for urban mobility issues, and even here, I’ve noticed today is Bike to Work Day, so you can use the public
bicycle infrastructure for free as a way to promote cycling. And as a result, more and more cyclists are present in the city. And we saw that as an
opportunity to engage them as potential ATI users. And cyclists are already
confronted with urban interfaces such as these bicycle-count displays that are present in the city. They show the number of
cyclists that pass by. But they’re also a good
way of involving cyclists, or give them the feeling of
being part of a community. Now, these cycle counters,
they capture more data than just the number of passersby. They also capture how
fast the cyclist is going, what type of cyclist,
the weather at the time, and our local government, which is the city of Leuven in Belgium, they asked us if we could
design something that would open up all these data
in the pubic sphere. So this was our research question. How can cyclists be informed
by an urban visualization? And then we had a second… We saw an opportunity to
further exploit this idea and look at the urban
visualization as a polling device. So our second question
was how can cyclists efficiently participate in public polling? Which is more an emphasis on interaction. Now, our local government
also had other requirements for the system, so they
really wanted to employ it. It needed to be mobile, and
therefore, it also had to be robust because it could not be anchored in the ground, this displayed. And it also needed to be affordable. We had our own requirements, that was that it should
allow or should show multiple data dimensions, and therefore, we designed it for cyclists
in waiting positions, for instance, at an intersection. And it should also allow
social interaction. And we looked at existing
work amongst others, this Looking Glass project, Where they looked at the
communicating interactivity towards pedestrians, and towards cyclists, we looked at Open Heart
Helmet project, for instance, where they visualized the heart rate data of the cyclist and communicated
to its environments and to its fellow cyclists. And for the public polling
and towards public polling for civic engagement, we
looked at Vote as You Go and Visualising Mill Road. So this is an overview of our
design and evaluation process, which I will now deepen into. So we started with
observations of cyclists in waiting positions at
intersections, and we noticed how they used their hands
or their feet to rest. This is also something we
saw in already existing urban furnitures such as
this handrail and footrest that’s being installed in Denmark. And we took this idea and
did some quick exploratory design studies with them. So this is at our campus, a bicycle lane, and we asked random
passersby, random cyclists, to stop at this pole and read the question on the paper display. And it was a yes or no question,
which we asked to answer it with their hands on this
handrail or with their feet on this footrest, while we
observed how they behaved. We saw that they were
wiggling with their wheel, so this inspired us also
to design a third method, which is using the wheel as an input. So we further developed
this into low-tech markups, and this was the evaluation
setup, so on one hand, we had a computer screen with
the polling question on it. On the other hand, we
had our three markups, other side of the intersection,
the other markups. And a researcher present
with a Wizard of Oz remote. For each interaction method,
we stopped 30 cyclists, and they had to fill in a short survey. And we also did a semi-structured
interview with them. And each of the interaction methods also had two alternative designs. So for the handrail, we had
a design with large buttons you could grab with your whole hands or one with small buttons
you could push with a finger. We had a vertical layout and a horizontal layout for the footrests. We had a layout for the floor
mat that you needed to go back and forth with your wheel, or the other one, on the right is when you needed to choose a
compartment to give your vote. So the outcome is that the footrests was rated as very uncomfortable. People not liked to gaze up and down, but the handrail was comfortable, and also they provided
reliable feedback, people said. But we also noticed, how the
floor mat was very attractive, and fun, and people were
eager to participate because of they were
curious of those floor mats. So we decided to combine those. For the LED display,
for the visualization, we did a code design workshop
together with cyclists and local governments. And we offered them all
the bicycle time data, and we were also open
for their data needs. And this led us to design
a kind of a story design for the LED display. So that started with the frustration, so those cyclists had
several frustrations. One of them was that
some other, some cyclists drive dangerously fast,
brushing other cyclists, crossing over intersections
without looking, giving them “normal” cyclists a bad name. So this led us to come
up with this question. What do you think of fast cyclists? For polling. And it was also shown in the data, so the top speed is much
higher than average speed, which supported this frustration. Now about implementation,
so this was our final, how our final prototype looked like, pressure-sensitive floor
mat, which was connected to a button on the handrail for support and an LED display where
instructions were put below. So how to use the system
were just written underneath. Now it was a challenge
to put a visualization on such a low-resolution screen. So we divided…so this is a bar graph. We divided it into intersections,
and you could navigate through the bar graph with your wheel. And next up is the top
versus average speed. And this one was a line graph also, and you could explore using your wheel. So these are the days,
Saturday, Sunday, Monday. After a one-day pilot study,
we did an in-the-wild study, and we did it at two
locations because we wanted to evaluate the impact of the environments. So we did the first two days,
we did a setup in the city, and the second, or the
third day, we did a setup in a suburban area for
three days, we did that. So at both locations, it was
placed at an intersection. All interactions were logged. We also put a video camera
in a car that was present, and it recorded all
activity around the setup. And the researcher was
there to do interviews with participants but also with people who hinted towards participation. So in total, we did 63 interviews. And based upon the video,
we logged all behavior and categorized it according to this model of Memoravic and his
colleagues, dividing it to passive and active
engagement over discovery. What was most noticeable was
that only 13% of all passersby completely ignored the installation. So the eight design recommendations, I will just discuss them briefly. We had several teasers that gradually announced the interactivity, but we noticed that, if we
learned that the combination of this passive teaser
with an interactive teaser, cyclists that drove over this floor mat, the screen would react, that this combination worked best. Then, in addition to the
existing landing zone that Muller and his colleagues described, we learned that cyclists
halted or stopped meters before they saw the installation
in order to be able to inspect what was going on. So we called this the
situational overview. Then using this familiar
interaction method first, a button, gives users confidence
that the system works, but also that they work,
that they know how to use it, which was a good introduction for our less familiar methods, we learned. Then, although the floor mat
was maybe less intuitive, people, cyclists found it
was designed specifically for them, and it made
them eager to participate. It was considered to be exclusive. But not only the physical
qualities of the installation appealed to them. Also, the relevancy of the
contents, in this case, in the suburbs, it was a
dangerous intersection, so people, cyclists wanted
to participate because they wanted to give their
opinion on this intersection. We saw the participation rate
in the city was much less, and we expect this has
something to do with the fact that this intersection is
perceived less dangerous, but also, we suspect that
this has something to do with the fact that cyclists
in the city do shorter trips, and thus have less time. So the location of this
visualization in a bicycle lane already gives meaning
to the visualization. But we learned that
also the interactivity, the fact that the had to use the real, added to the sense-making process. And then lastly, the social interaction. Cyclists, they take up more space, so social interaction was often limited because they had to shout to each other. And we also saw how more users on one bike
could work together. Like this mother would steer her wheel, and then one of the children
would press the button, and they could play together. So for now, about the validation,
a more permanent prototype is being built by a
company in public display, namely Qlight, which is active in Belgium and the Netherlands. And it looks like this. We’re very curious to
try it out this summer. The province and the city
of Leuven will implement it. And we’re curious how
to validate our results. So in summary, we learned that
the high participation rate demonstrates that there is
this kind of societal demand for these kinds of interfaces. Second, that there are
opportunities to develop public interfaces for
specific user groups. And third, that such interfaces
not only should be or could be about polling but
also should be about informing. Okay, that was it. Thanks. (applause) – [Aaron] Hey, whoa, here you go, okay. So, how we’re going to work the questions during this session is the
SVs are kindly going to run around with the microphone. So, I’d like to actually ask
people to stay in their seats just to make it more accessible. So we don’t all have one little microphone that we all cue up to. So, you just put your hand. You don’t need to keep it up for long if you want to ask a question. And then I’ll send the
microphone in your direction. No? Any questions? I’m going to start off then. So in the paper itself,
you allude to this question of using alternative sensing technologies. So this question of let’s
say using computer vision, for example, so you put up your left hand to give one answer. You put up your right hand
to give another answer, and you cycle by… At 20, 30 kilometers, I have no idea. I actually don’t cycle
very much…miles an hour. Could you talk about that
because in your future work, you discuss other instanciations. So, why is the physicality so important, do you think? – Well actually, we wanted to
explore that option as well, like now it’s for cyclists
in waiting positions, and we wanted to explore it
while they were driving too. So which is more that they
could poll while cycling like in a more extensive way. And actually, we’re currently planning to set up research about
that on cycle highways because here, it’s a city,
and I think it’s important that it should look
integrated, more urban. Therefore, this looks more
like a traffic light design, urban infrastructure design,
so I think that’s important when you implement something
like this in the city. – [Aaron] So it’s integrated
into the landscape, the architectural landscape. – Yeah, yeah, yeah. – [Aaron] Okay. – So just difference in context. – [Aaron] Okay, so here we go. – [Man] Hi, Carl Gutwin from
University of Saskatchewan. I’m interested in the novelty effect. Since this was only
deployed for five days, and I just wonder what
you think would happen. Will people just start
to ride by after awhile? Or have you thought about
how do you keep them engaged? And how do you keep them stopping? I realize sometimes
they’re stopping anyways, but how do you keep them
looking at the display? – Yeah, the thing is, of course, we mention that in the paper too, the novelty effect, but we suspect that it’s a mobile installation,
so it will be to travel around the city. There’s several counters
throughout the city, and we will install it at several places, so it will always be
novel for some cyclists commuting to work. But also, the question changes. It’s not that it always
will show this question. You can adapt to other polling. So I guess, and as we
noticed too, that people are interested to give their opinion. Cyclists wanna give their opinion, so we think that’s the way to overcome it. – [Aaron] Okay, so do we
have the one last question before we thank our speaker for her talk? No? We’re all post-lunch blues, no is that it?