Governments, organisational institutions and tech companies around the world are not only collecting and analysing data, but increasingly turning to urban robots for automating processes and services, with the aim to make cities more efficient and productive. However, the development of so-called smart cities also triggered heated debates among thought leaders and scholars, with some arguing that the matter of productivity should be dissociated from speed and efficiency and instead questioning how to make cities more liveable and healthier, regarding people as co-creators instead of simply consumers, and create cities for slowing down.
In this exploratory design research project, we created Woodie, a slow-moving robot that draws on the ground using conventional chalk sticks. Capable of drawing various predefined designs, Woodie uses the public space as a large horizontal canvas, producing simple line drawings. Building on previous research on pervasive urban displays, we were interested how an urban robot could replicate the qualities of non-digital public displays and act as a facilitator or ‘spark’ to affect people’s engagement with public visualisations and to foster social interaction among people. As robots are increasingly being tested in real-world urban contexts — mainly to complete mundane tasks — our aim was to investigate their potential for triggering urban reflection, “getting lost” in cities and “slowing down”.
This article is a summary of a full paper submission accepted to the ACM CHI Conference on Human Factors in Computing Systems (CHI’20) with the title “Stop and Smell the Chalk Flowers: A Robotic Probe for Investigating Urban Interaction with Physicalised Displays”. The conference was supposed to be held between April 25–30, 2020 in Honolulu, Hawai’i, USA, but was cancelled due to the current Covid-19 pandemic. The paper is published in the ACM library (https://dl.acm.org/doi/abs/10.1145/3313831.3376676), and a full-text is available here.
Pervasive displays in the city
Our study builds on and contributes to the field of pervasive displays, which has been extensively studied by the CHI community — as a field that takes computing into the urban space through digital screens that can be either interactive or non-interactive. Compared to smartphones, pervasive displays enable a push-based distribution of content that is addressed to the public instead of individuals. However, several works reported that passers-by tend to ignore digital screens. Therefore, researchers investigated how to overcome so-called display blindness, for example, through adding layers of interactivity or aiming for the design of more relevant content.
Several studies also investigated and documented the use of non-digital displays, for example, through using chalk spray stencils to display urban visualisations on the pavement or deploying chalkboard-like signage for comparative energy feedback on the façades of residential houses (Figure 2). Based on their studies, the authors outlined several qualities of non-digital public visualisations: for example, they reported that textures and materiality, as well as the horizontal position of content on the pavement, attracted passer-by to approach (and touch) the visualisations. Further, the transient nature of the visualisations made people aware of their finite lifetime, resulting in additional appreciation of the content. The manual update process carried out by a research team created additional engagement between residents and the researchers acting as facilitators. Yet, although those studies suggest that non-digital physicalised displays can address some of the socio-technical pitfalls related to digital urban displays, their hyperlocal scale and impact remain unaddressed limitations.
From the outset, the design goal was to create a robotic device which could render digital drawings in a physicalised form, deeply integrated into the existing urban environment. One requirement was to design a ‘plug-in system’ that could be deployed “anywhere” in public space, in and out, without the need for additional infrastructural support, such as a canvas or a stationary power source. As sidewalks and public plazas are ubiquitous in cities, publicly owned and used as a stage for social interactions, we decided for a self-moving, autonomously powered platform which uses the ground as a large canvas.
Shape. In line with our goal of enticing curiosity by depriving Woodie from any obvious association with preconceptions about how a robot should look like, we decided early on not to give it anthropomorphic features (Figure 3). Instead, we opted for a circular body, thus allowing the robot to move in any direction without the constraints of having to reposition its “face”.
Size. Woodie should be large enough to be noticed by passersby, yet small enough not to become an obstacle to them, and not to obstruct visibility of its own drawings.
Speed. Slowness emerged as an important feature for Woodie, as it should move sufficiently slow to allow for thoughtful appreciation by the public, as well as not to compromise the free movement of pedestrians in the public space.
Information visualisation. The information visualised through the robot’s drawings — representing an instance of the physicalised display — had to be aligned with the environment and its physical and temporal context. Given that Woodie was located in a laneway, we aimed to use visualisations for activating the space, implementing digital placemaking principles to connect people with the space and with each other. As we deployed Woodie in the context of a public light festival, we aimed to align the visualisations with the theme of the festival, which was “love, peace and harmony”. These considerations led us to the use of simple shapes, such as flowers and hearts. An important aspect at this prototyping stage was not only to decide for the “right” type of content, but also to explore the characteristic style of the drawings. After several tests drawing on various grounds, we realised that highly geometric vector graphics convey the impression that the drawings are imperfect when “rendered” on rough terrain, while less geometric, hand-sketched drawing styles worked well with Woodie’s limited drawing accuracy (Figure 4).
Information communication. In addition to creating a public display through drawing on the ground, Woodie should also have communicative features itself. Following recommendations from research on autonomous vehicles, suggesting the use of visual signals for communicating a vehicle’s intent to people around the vehicle, Woodie should be capable of visually representing its internal mode. We chose a low-resolution LED lighting display solution given their established aesthetic qualities.
Deploying Woodie in-the-wild
We deployed Woodie over the period of three weeks, during evening hours, in a quiet laneway situated within a major business and residential district. Due to the slow movement, Woodie was usually not drawing more than four individual drawings per hour.
Types of engagement
Based on our observations and interviews, we identified several types of engagement: when approaching the location and taking the decision to stop by and contemplate, people were mainly curious and started speculating about what “the robot is doing”, the drawings on the ground, and other people interacting with and around Woodie. When recognising Woodie as a robot, which was mostly based on the movement (“I saw the movement, and could tell it’s a robot”), people started to speculate on its task. Relating to the drawings on the ground and its resemblance with the home cleaning robot Roomba, many people thought in the beginning that Woodie would be “wiping the floor”, so that “other people could draw again”.
If the intervention successfully evoked people’s attention, then this initial state of being curious would transition into a phase of observing: people who were more interested on the performance itself, started “walking around, and observe [Woodie] from different angles”. At this stage, people often realised that Woodie was drawing, which was sometimes not immediately visible due to the prominent drawings made by humans. Interviews confirmed our observations from onsite visits and video surveys, that people often stayed at least until a drawing was done, in order to see “how long will it take for [Woodie] to finish a drawing”. Recording the drawing process with a smartphone additionally enticed people to wait until the drawing was completed.
The opportunity to manually draw on the ground successfully engaged people in actively participating in our intervention. People appreciated the “feeling of belonging” and that “[the intervention] includes everyone”, referring to the lane as a “graffiti place that makes you feel relaxed”. While drawing on the ground was particularly popular among children, we also observed many adults adding drawings. Actively participating in the intervention was also another reason why passers-by ended up dwelling rather than rushing away from the precinct, further exploring the scope of possibilities offered by the intervention. One interviewee who stated to have spent more than 20 minutes at the site, reasoned: “You get to look around at everything, draw, add to things and see what other people are doing.” — which is evidence of the intervention achieving its aim of connecting people with the space and with each other. Importantly, this retention of passers-by in the space seemed to be driven not by any of the isolated design elements — e.g. Woodie’s presence alone, or the availability of chalk sticks to draw on the pavement — but rather by the possibility of engage in a playful activity, in public, facilitated by an urban robot:
“I like the fact that it’s kind of a mixture of something, let’s say, non-tangible and tangible. So the combination of we can draw ourselves, but the robot draws as well, is a nice touch. It makes it more human, I suppose, than only a robot [drawing].”
Audience behaviour and grouping
When successfully engaged in longer-lasting engagement, we observed that some people, in particular children, tended to get very close to the robot, while others would keep a distance because they were “worried to disturb Woodie” and thus risk not being able to see the finished drawing. When observing Woodie, children, but also adults, would often bend down, or sit on the floor, in order to understand the mechanism whereby the chalk stick touched the ground. Interestingly, despite Woodie’s spherical shape, looking similar from all sides, people were often readjusting their standing position to see Woodie from various sides. One explanation for this behaviour could be that, depending on which direction Woodie was drawing, the chalk stick which was pulled behind could be more visible and better observed.
In this longer-lasting engagement situations, we could observe two types of audience groupings, depending on the availability of chalk sticks for the public: in situations when chalk sticks were not available, and therefore only Woodie was drawing, we observed focused groups of people standing around it as if it were a street performer (Figure 6, left). Conversely, on the majority of evenings, when we did provide chalk sticks to the public, we observed a diffused distribution of people spread over the whole area (Figure 6, right). This diffused distribution of people was enticed by what we refer to in our paper as the “Cracked Honeypot Effect” (section “Types of Engagement”, page 7): the emerging content created new display spaces, at which people stopped by, reflected on the existing drawings and added their own content. On the contrary, later in the evening, when Woodie’s allocated zone was covered by drawings, people had to look after ‘free’ display spaces, which either resulted in people drawing over existing content or ‘expanding’ the canvas towards the adjoining street.
Our data analysis revealed that people had a variety of emotional responses to Woodie. Many referred to Woodie as something “cute”, that they want “to take home”. One woman who performed in front of the robot — putting her hat on it — mentioned to the group of people around her: “it’s not interactive but it makes people interact, it responds to me like nature responds to me.” The pattern of relating Woodie to natural phenomena was also apparent in comments by other interviewees. For example, one female interviewee stated that the robot “feels more like a living organism”. While some people associated the robot to known living species on earth, e.g. a “mushroom” or “jellyfish”, others rather perceived Woodie as a creature “from outer space”. Sometimes people were also speculating about gender and character traits: for example, one elderly man, after observing Woodie for around 20 minutes, asked whether Woodie is a “male or a female” and if it’s “serious”.
Using the ground as a canvas. Rendering visualisations on the ground attracts people, as the content becomes visible in their immediate physical environment. A visible rendering device moving on the ground, in the form of an urban robot, entices people to shift their gaze to the floor and pay attention in order to adjust their walking accordingly, which in turn causes them, at least momentarily, to slow down. Using an urban robot to print content in a physicalised form allows to arbitrarily increase the ’display area’, which is merely restricted by physical rather than technical constraints. As a consequence, the “catchment” area for passers-by spreads way beyond the robot itself, and in spite of it.
Urban robots as a facilitator. For making people dwelling on a space they would otherwise just have walked past, physicalised urban displays could also become a powerful strategy for placemaking and community engagement. In that sense, we would argue that Woodie successfully acted as a facilitator for participatory, creative placemaking. Firstly, the integrated low-res lighting display attracted people to approach the intervention site and explore the chalk drawings on the ground. Secondly, the curiosity triggered by Woodie’s presence, fostered social interactions among people, such as talking about the robot and the content itself. Previous work reported that researchers manually updating content functions as a “spark” for social interaction, which is nonexistent with digital public displays. Therefore, we argue that urban robots, visibly creating physicalised displays, can replicate this role.
Slow pace urban media. Early studies reported that delayed, periodic update cycles can create additional anticipation around urban data visualisations. Slowing down the rendering process itself, we found can further increase the dwell time of people and entice them to “come back to see the finished drawing”. Future research needs to investigate the influence of speed levels on the interventions’ temporal trajectories, such as people’s disengagement and reengagement with it.
Characteristic style. Our study confirmed the attractiveness of non-digital physical displays in regard to their materiality and transient qualities (e.g. people touching the chalk). Additionally, the characteristic aesthetics of the robot’s drawings in comparison to those made by people added another layer of implicit information to the content (e.g. “We saw similar, precise chalk drawings, so we realised it must be the robot”). Further, the unpredictability of movement due to the surface of the ground can add analogue aesthetics, however needs to be considered in the chosen drawing style.
The authors of this article are members of the Design Discipline at the University of Sydney School of Architecture, Design and Planning.