Roam.io

How can we enrich sensor-based data with human input?

Partners

UCLIC, Madeira Interactive Technologies Institute (M-ITI)

Our ‘in the wild’ study demonstrated how Roam.io successfully engaged the public to answer a range of questions and interpret data visualisations.

Opening data by enabling the public to comment on and respond to sensed IoT data (in our case, WiFi activity data) has great potential to enable the public to perceive and understand data about themselves and the environment while also providing new insights that can inform the management of public services.

Newly emerging urban IoT infrastructures are enabling novel ways of sensing how urban spaces are being used by people. However, the data produced by these systems is largely context-agnostic, making it difficult to discern the meaning of patterns and anomalies in the data.

To address this challenge, we developed a hybrid approach that combines quantitative data collection through an urban IoT sensing infrastructure with qualitative data collection from people answering relevant questions in situ. For this purpose, we developed a new way to survey the public through a robot-like, physical installation called Roam.io.

Roam.io is designed to encourage the general public to engage with urban IoT data through data visualisations of sensor-derived data and relevant questions. It allows them to voice their opinions, comments, and perspectives on the data displayed. A hybrid dataset is created from the combination of the quantitative sensor data with people’s qualitative in situ responses and contextual observations. This dataset uniquely helps to build a better understanding of what is happening in this urban space and allows for new insights from the gathered IoT data.

The goals of Roam.io were to:

entice passers-by to walk up and interact with a public physical installation

engage the public in data exploration through visualisations and interactive questions

see whether the public can help explain anomalies and changes in the data

explore whether the public is willing to complement data logs with subjective interpretations, opinions, and in situ observations

Roam.io was deployed in the city centre of Madeira, Portugal as part of a week-long study. Madeira is a popular tourist destination, seeing as many as 1.2 million visitors each year. With a population of only 270,000 people, local authorities have become increasingly concerned about the

economic, ecological, and social impact of tourism on the island. To track where tourists visit and how many go where, an infrastructure has been set up throughout the island that provides time- and place-sensitive people count data (or ‘people flows’) using passive Wi-Fi hotspot analysis.

We designed the Roam.io installation with a subtle anthropomorphised robot look to create a friendly and open interface for people to approach in a public setting. Roam.io asked a broad set of 34 questions. These included 5 demographic questions about nationality or language, 8 contextual questions asking the user to describe the environment, 10 data questions in which users were asked to comment on a data infographic, and 11 factual questions about Madeira Island.

The visualisations depicted historical tracking data across days, weeks, and even months. Each question asked users to ‘vote’ for the best matching interpretation or contribute their own interpretation of the dataset via the installation’s keyboard. The visualisations were designed as simplified graphs that could be easily read, showing peaks and valleys in people flows at different locations and times on the island. We were interested in finding out whether people could infer what these represented; for example, whether a peak on a Saturday afternoon indicated an influx of people on a tour visiting a tourist site in the city centre or locals attending a festival. This hybrid data was then shared with the tourist board and other communities to help them develop a better understanding of the impact of the people flows on specific areas.