Walking with Geo-Zombie: a Pervasive Game to Engage People in Urban Crowdsourcing

Transcription

1 2016 International Workshop on Networking Issues in Multimedia Entertainment, ICNC Workshop Walking with Geo-Zombie: a Pervasive Game to Engage People in Urban Crowdsourcing Catia Prandi, Paola Salomoni, Marco Roccetti Dept. of Computer Science and Engineering University of Bologna Bologna, Italy {catia.prandi2,paola.salomoni,marco.roccetti}@unibo.it Valentina Nisi, Nuno Jardim Nunes Madeira Interactive Technologies Institute University of Madeira Funchal, Madeira {njn,valentine}@uma.pt Abstract In this paper we present results gathered from field trials while using a pervasive game, called Geo-Zombie. It intermixes reality with virtual zombies, providing a game experience to be lived in the urban environment. Geo-Zombie is designed for a specific goal: engaging people in collecting georeferenced data about urban accessibility. The game has been designed with the aim to involve a large number of players/volunteers in the activity of signaling urban barriers while walking. To get ammo to react to a zombie attack, in fact, players can sense/map (crowdsensing) urban barriers and facilities and transmit those information back to an operative center (crowdsourcing). With the aim to assess the efficacy of Geo-Zombie, we conducted real experiments contrasting three different mobile apps, specifically designed to: i) simply collect urban information without providing any gaming experience (Basic), ii) reward volunteers that participate in the crowdsourcing activity (Reward), and iii) provide fun and entertainment to players who contribute (Geo-Zombie). We here provide results along two different perspectives: a quantitative one (e.g., number of collected data and similar) and a qualitative one (i.e., the players experience). These results confirm the feasibility and suitability of our approach and stimulate interesting discussions. Keywords pervasive game; crowdsourcing; urban field trials with results; accessibility I. INTRODUCTION As the global population continues to grow, the number of people moving and living in cities continues to increment. To deal with this situation, the concept of smart city emerged as one of the primary factors to investigate and enforce. A smart city can be defined as a city well performing in a forward-looking way in economy, people, governance, mobility environment, and living, built on the smart combination of endowments and activities of self-decisive, independent and aware citizens [1]. IT infrastructures and emerging ICTs are enablers for smart cities, redesigning our urban environments and becoming more and more pervasive thanks to the diffusion of ultra-low power sensors, wireless networks, powerful data connections, and web and mobile-based applications. The new paradigm of IoT (Internet of Things) is emerging as one of the major technology trends that will have a tremendous impact on life, connecting people, data, things and process [2]. In this context, also the concept of game is changing in order to fit the huge diffusion of smart devices and of the IoT paradigm. Games are becoming more pervasive and ubiquitous: they provide players with novel captivating and sometimes mesmerizing ways to enrich the player experience, intermixing reality and virtual in a new immersive scenario [3]. Along this line, the contribution of this paper is to show and discuss results obtained from field trials with a pervasive game, called Geo-Zombie, designed and developed to exploit the game flow with the aim of involving people in mapping data, while escaping from zombies. The collected data will be used by mpass (mobile Pervasive Accessibility Social System), a way-finding system, that provides personalized and accessible paths using also data obtained via crowdsensing [4]. In order to evaluate our approach in using a pervasive game in a crowdsourcing context, we implemented other two mobile apps to compare the obtained field trials results with Geo-Zombie. The two apps are: the Basic app, it is the simplest one and it just allows users to collect data; the Reward app that applies gamification elements and external rewards to involve volunteers in crowdsourcing data. We tested these two apps with the aim of contrasting the Geo-Zombie approach in gathering data about the urban environment. The peculiarity of our work is that we investigated the game flow of a pervasive game to engage the largest number of players in collecting valid data via crowdsourcing. In fact, the final aim of the game is to increment the quantity and the quality of data, so as to achieve the critical mass needed by a specific crowdsourcing system to provide effective services about urban accessibility. We analyzed the results of the field trials both from a quantitative viewpoint (e.g., number of collected data and similar) and from a qualitative perspective (i.e., the players experience), comparing the findings of the Basic app and the Reward one with Geo-Zombie. The outcomes show the feasibility and suitability of our approach in using Geo-Zombie, generating interesting discussions. The remainder of this paper is organized as follows. In Section II we briefly survey crowdsourcing systems that offer services about urban accessibility and we provide the background about the whole mpass system to better understand the research challenges related to this paper. In /16/$ IEEE

2 Sections III we describe the design and the implementation of Geo-Zombie. Section IV illustrates our experimental tests and the results we obtained by contrasting and comparing Geo-Zombie with the other two mobile apps: the Basic and the Reward ones. Finally, Section VI concludes the paper. II. BACKGROUND AND RELATED WORK In the last few years, different services have been developed with the purpose of mapping data about urban accessibility, exploiting crowdsourcing. Some examples of these systems are: Wheelmap [5], AccessToghether [6], and AXSmap [7]. These systems focus on collecting information about the accessibility of points of interests. Others systems collect data via crowdsourcing with the aim to provide accessible paths. Some examples are: OpenTripPlanner [8], and Path 2.0 [9, 10]. These systems provide personalized routing to mobility impaired pedestrians and wheelchair users, but they fail in achieving the critical mass needed to provide effective services, due to the difficulty in collecting a large amount of trusted and updated data. To increment the quantity (amount) and the quality (trustworthiness) of urban accessibility data, we developed mpass (mobile Pervasive Accessibility Social System), a system that allow citizens to collect data related to the urban accessibility [4]. In particular, via crowdsourcing and crowdsensing, people can participate in mapping barriers and facilities in the urban environment using a context-aware mobile app [11]. The final goal of mpass is to provide an accessible way-finding system, computing accessible and personalized paths according with users needs and preferences [12]. To increment the quantity of information, our system integrates data provided by crowdsourcing, crowdsensing and official data. The consequence is an important increment in the amount of collected data. Despite this, the amount of gathered data is not enough to provide an effective service due to the peculiarity of such data. In fact, the most of contributors in the system are people with special needs or preferences. Often, for other people is difficult to face with the importance of collecting data about barriers and facilities, and, thus, with the need of personalized paths. Some crowdsourcing systems exploited game mechanisms and game design elements to engage and involve people in proving data [13, 14]. A famous example is the ESP game [15], a simple online two-player game in which the goal is to guess what label the other player would give to the displayed image. These labels will be used to tag images, with the aim of improving the accessibility of images for visually impaired and blind people. ESP is the first example of game with a purpose (GWAP), a game in which people, as a side effect of playing, perform some useful tasks in a crowdsourcing way [16]. Along this idea of exploiting game thinking and game design in involving people in crowdsourcing activity, we have designed and developed our mobile apps. III. MOBILE APPS IMPLEMENTATION We adopted an iterative design process in order to conceive, refine and prototype the mobile apps [17]. This process included: (i) the ideation and the sketch of the game Fig. 1. Geo-Zombie: a photo of the experience prototyping and a app screenshot. concept; (ii) the refinement through a series of feedback sessions and focus groups; (iii) the simulation of the game flow by experience prototyping sessions. According to the results of the design phase, we developed the Basic app, the Reward app and Geo-Zombie as Phone-Gap ( apps that can run for Android, iphone, Windows Phone and other platforms. We tested the apps on different smart devices running Android 4 (or higher) and ios 7 (or higher). Those smart devices feature a touch screen, a build-in GPS receiver, a camera, an accelerometer, a gyroscopic, and a Wi-Fi or cellular data connection capabilities. We wrote the software by using web technologies (HTML, CSS and JavaScript) and the Cordova plugin APIs ( We used Google Maps JavaScript API v3 ( to manage the position of the user (and zombies). The back-end part has been implemented as a RESTFull Web Services using Java 7 and the Spring framework ( Data are stored in a PostgresSQL DB ( integrated with PostGIS ( to enable support for geographic objects. The Basic app is a very simple app created to collect data about facilities/barriers in the urban environment. Using such an app, the user can easily map urban elements, sending information about the GSP location and the typology. In the Reward app, we have applied some gamification elements to involve people. In particular, the app exploits extrinsic motivations expressed in terms of a concrete reward (a voucher) to engage users in gathering data. The app uses the GPS position and the user s declared interests to look for nearby stores or markets that offer vouchers. The more the user collects data via crowdsensing and/or crowdsourcing, the more he/she obtains pieces of a voucher. The goal is to obtain all the pieces of a voucher, like in a puzzle. In the Geo-Zombie game, we use the camera and the direction of the phone, obtained through the gyroscope, to display zombies in the real environment. The core of this app is the Game package that is composed by two main modules: (1) PLAY, and (2) MAP.

3 PLAY: allows people to start the round and visualize the current map. To begin the round, the system has to follows 3 steps: (1) to obtain the GPS position of the player; (2) to randomly create and locate zombies around the player (with a maximum distance of one kilometer), and (3) to orchestrate zombies using GIS-based routes, by setting as their walk destination the player's GPS location. The marker position is refreshed during the game according with the current updated user GPS location. Zombies walk at different speeds with the aim of catching the player, following the shortest path to reach his/her position. When zombies invade the area around the user (in an around of ten meters), the system alerts the player and switches the view from the map to reality (camera visualization) and the player has to flush out and shoot zombies, to prevent the zombie bite attack. Fig. 1 compares a picture taken during the experience prototype (the left one) and a screenshot of the Geo-Zombie app showing the camera view (the right one). The round finishes when the player reaches the declared destination. MAP: permits people to document the urban accessibility environment, collecting data about barriers/facilities. Through this functionality, people gain points to spend for ammunitions, weapons, and energy. To map an element, players need to choose the corresponding category of the urban elements from a defined list. Moreover, they can insert additional information (like description, notes and photos) and then obtain more points. These data will be sent to the server, together with the current GPS position. During the mapping activity, zombies are stopped to facilitate - and not penalize - the player. The map activity can be a voluntary action (the player decides to document the urban element) or it can be the result of a system notification to confirm the presence/absence of a close barrier/facility. The aim of this feature is validating collected data; it can be used together with the sensing activity and the tracking of user's paths to reinforce the quality of data and punished cheater players. The system follows the user's trace during the round: the more the player walks, the more gets points as reward of the movement activity. The track activity is important to infer the trustworthiness of collected data, comparing information related to the same path and gathered from different users. Another mechanism implemented to limit cheating mapping is related to the points assignment: the more different players document the same barrier/facility, the more points they gain. IV. FIELD TRIALS AND RESULTS To evaluate the feasibility of our approach, we conducted field trials comparing Geo-Zombie with the other two mobile apps we developed: the Basic app and the Reward app. We asked a class of fifty students to experience all the three apps. Such students are enrolled on the Multimedia System Technologies class of the Bachelor Degree Course in Computer Science and Engineering (University of Bologna, Italy). A first element of interest is the fact that almost all the students accepted our proposal and took the decision to participate (almost 95%). We decided to recruit students in field trials who denote a target of avid players and daily walkers. TABLE I. QUANTITATIVE RESULTS Apps N. of users N. of report N. of report per user (Average) Basic 48/ Reward 47/ Geo-Zombie 48/ Fig. 2. Answers to the item The app was boring. N. of report per user (Standard deviation) The students attend courses in Cesena, but most of them live in different cities, including Bologna, Forlì, Faenza, Rimini, and Ravenna. These cities have different densities and layouts, relevant factors in examining the variety of urban environments. We studied the use of the Basic app, the Reward app, and Geo-Zombie the students have done over three weeks (a week per app). The first week they could mapping barriers using the Basic app; the second one, they could obtain a voucher with the Reward app, and, the last week they managed to fight with zombies by using Geo-Zombie. We made the apps available to the students and we monitored their use by exploiting logging data in our DB. After each field trial week, we involved students in filling a survey related to the app of the week. The survey was composed of 16 items: the first 15 questions were ranked using the Likert scale approach based on a symmetric agree-disagree scale [18]; the 16th item is an open question to insert personal comments. We analyzed resulted data from (i) a quantitative viewpoint and from (ii) a qualitative perspective, obtaining interesting results. A. Quantitative results Data related to quantitate results are shown in Table 1. Using the Basic app, only 24 out of the 50 participants did more than a report, which was required to start the app. In the Reward app, we asked the students to voluntarily report at least 5 apois, to get the voucher (which corresponds to a class assignment marked as done). 40 out of the 50 students completed the assignment, 22 of which did more reports than the 5 ones required. The total amount of reports done was 3.3 times the ones obtained with the first app. With Geo-Zombies,

4 34 out of the 50 did more than a report, which was required to start the game, but the total amount of reports done was 3 times the ones obtained with the first app. Fig. 3. Answers to the item I come out from my daily paths. Fig. 4. Answers to the item I m more aware with the presence of barriers/facilities. In summary, using the Basic app on its own we got the lowest number of reports per person per week and the lowest interest in participating in mapping the city. With the Reward app there was a significantly increase of the number of reports per student due to the external motivation (the voucher). In fact, the average number of reports for person is very close to the number of report required to obtain a class assignment as done. Playing with Geo-Zombies, we had a valuable increase in the number of reports. In particular, from the high value of standard deviation (16.60), it is possible to infer that the more the player enjoys the game flow, the more he/she is engaged in the mapping activity [19]. B. Qualitative results For the qualitative perspective, we analyzed the students answers in the surveys. Each survey was composed by 5 items related to the goals of the system, 10 items connected to the specific tried app, and one open question for comments and hints. Some interesting results emerged from the surveys, related to different investigated issues. The first issue emerged from the surveys was the relation between the enjoyment in using the apps and the engagement in the mapping activity. Fig. 2 shows the students answers to the item The app was boring. As expected, comparing the results obtained from the three apps, the Basic app was considered by the most of students as the more boring. Interesting results are related to the Reward app if integrated with the quantitative data. The app is considered neutral (neither boring nor funny) by the 42% of the students, but this feeling didn t compromise their engagement in reaching the voucher (valid class assignment). For example, Matteo affirmed: I found the Reward app more engaging that the Basic one, it wasn t funny, but I wanted the reward so I continued to report elements until I have achieved it! Data in Fig. 3 show an interesting result strictly related with the feasibility of the system in reaching the critical mass needed by mpass to become effective. With the Basic app, students are not interested in changing their routine and their daily paths. Different results are obtained with the Reward app and with Geo-Zombie. In particular, it is very impressive that, by using Geo-Zombie, more than 50% of the students (22% strongly agree and 32% agree) explored new paths, broking their routine. This is due to the game flow that allows players to forget about their daily habits in order to enjoy the game. For the first time in three years I took a different route to reach the railway station as I had Zombies behind me, and I discovered a very interesting route, full of small food shops!! Another important aspect highlighted by the results, is related to the awareness of the urban environment. As shown in Fig. 4, the Geo-Zombie app and the Reward app enforce the awareness of barriers/facilities in the urban environment. Data show a correlation between the number of reports done and the awareness reached: (i) the Basic app obtained 1.97 average reports and an awareness (both Agree and Strongly agree answers) increased of 26%; the Reward app got 6.60 average reports and an awareness increase of 51%, and, finally, (iii) the Geo-Zombie app obtained 5.95 average reports and an awareness increase of 51%. It is worth noting that for the Geo- Zombie app, the Strongly agree answers are about 19% while they are about 15% in the Reward app, witnessing an higher level of engagement in the Geo-Zombie game. This emerged also from students comments. For example, Maria wrote: Using the Reward app, I found a lot of unexpected urban elements that give problems to people with disabilities; just in the area which I supposed to know well! Instead, Luca maintained: I founded the zombies idea very cool but I don t want run away from zombies during my daily activities I don t want run at all! C. Geo-Zombie Evaluation Analyzing students answers strictly related to Geo-Zombie, we found some interesting results that prove the feasibility of our approach in exploiting the game flow to engage users. By using Geo-Zombie, the most of the players (63%) declared they have just focused on gain points to buy munitions/guns with the aim of killing zombies (as show in Fig. 5). Ilaria maintained: At a certain point, I was so worried about zombies that I started to walk quickly in the opposite direction to

5 find news barriers! I really needed points to buy munitions... Zombies were so close to me!. ACKNOWLEDGMENT Our thanks to the fifty students of the Computer Science Degree of the University of Bologna (Cesena Campus) who took part to the experiments described in this paper. Fig. 5. Answers to the item I just wanted to kill zombies. Fig. 6. Answers to the item I tried to kill the largest number of zombies. This condition is the result of the game flow created by the pervasive game that allows experiencing it in a more immersive way. Another relevant aspect is highlighted in Fig. 6. The 47% of students have been pushed by the game flow in killing as many zombies as possible. Maximize the enjoyment for this kind of players is a key factor in incrementing the number of mapped accessibility urban elements. Anita, wrote: I gained a lot of munitions in my neighborhood just to walk versus the closer zombies and kill all of them! It was really funny! V. CONCLUSION This paper presents field trials and related results about Geo-Zombie, a pervasive game designed and implemented to involve and engage people in crowdsourcing accessibility urban elements. We conducted the experiment involving fifty students of the Computer Science Degree of the University of Bologna. We asked the students to experience three different apps in order to compare the obtained results: a Basic one, a Reward one and Geo-Zombie. The resulted data have been analyzed both from a quantitative viewpoint (e.g., number of collected data and similar) and from a qualitative perspective (i.e., the players experience). The field trials results show that the game flow is the key factor to investigate to captivate users in crowdsourcing mapping activities. REFERENCES [1] R. Giffinger, C. Fertner, H. Kramar, R. Kalasek, N. Pichler-Milanovic, and E. Meijers, Smart Cities: Ranking of European Medium-Sized Cities. Centre of Regional Science (SRF), Obtober [2] O. Etzion, F. Fournier, and S. Arcushin, Tutorial on the internet of everything, in Proc. of the 8th ACM International Conference on Distributed Event-Based Systems (DEBS '14). ACM, New York, NY, USA, , May [3] C. Magerkurth, A. David Cheok, R. L. Mandryk, and T. Nilsen, Pervasive games: bringing computer entertainment back to the real world. Comput. Entertain. 3, 3, 4-4. July [4] C. Prandi, P. Salomoni, and S. Mirri, "mpass: Integrating people sensing and crowdsourcing to map urban accessibility," in Proc. of the 11th Consumer Communications and Networking Conference (CCNC), IEEE, pp.591,595, Jan [5] Wheelmap, July 2015, [retrieved: July, 2015] [6] Access Together, [retrieved: July, 2015] [7] AXSmap, [retrieved: July, 2015]. [8] OpenTripPlanner, [retrieved: July, 2015] [9] M. Roccetti, G. Marfia and C.E. Palazzi, Entertainment beyond divertissment: using computer games for city road accessibility. Comput. Entertain. 9, 1 9, [10] C. E. Palazzi, L. Teodori and M. Roccetti, "Path 2.0: A participatory system for the generation of accessible, in Proc. of the International Conference in Multimedia and Expo (ICME), IEEE, pp.1707,1711, July [11] C.G. Harris, The Application of Crowdsourcing and Games to Information Retrieval, Bulletin of IEEE Technical Committee on Digital Libraries, Vol. 8 Issue 2, [12] C.E. Palazzi, M. Roccetti and G. Marfia, Realizing the Unexploited Potential of Games on Serious Challenges, ACM Computers in Entertainment, 8, 4, [13] S. Mirri, C. Prandi and P. Salomoni, "A Context-Aware System for Personalized and Accessible Pedestrian Paths", in Proc. of the 2nd International Workshop on Location-based Services and Applications in Ubiquitous Computing (LSAUC 2014), HCPS 2014, IEEE, July [14] S. Mirri, C. Prandi, P. Salomoni, F. Callegati and A. Campi, "On Combining Crowdsourcing, Sensing and Open Data for an Accessible Smart City", in Proc. of the 3rd International Conference on Technologies and Applications for Smart Cities (I-TASC), IEEE, September [15] Ahn, L.V. and Dabbish, L. Labeling images with a computer game. in Proc. of CHI '04. ACM Press, , [16] Ahn, L.V., Games with a purpose, Computer, vol.39, no.6, 92,94, [17] C. Prandi, V. Nisi, P. Salomoni and N.J. Nunes, From gamification to pervasive game in mapping urban accessibility, in Press. of the 11th Edition of CHItaly, the biannual Conference of the Italian SIGCHI Chapter, [18] D. Bertram. Likert Scales are the meaning of life. Available from: [retrieved: June, 2015] [19] P. Salomoni, C. Prandi, M. Roccetti, V. Nisi and N.J. Nunes, Crowdsourcing Urban Accessibility: Some Preliminary Experiences with Results, Accepted in Press. of the 11th Edition of CHItaly, the biannual Conference of the Italian SIGCHI Chapter, 2015.