Pervasive Game Development Today

March 2005
Author: Fabien Girardin

Introduction

In this document, I talk about my experience in developing CatchBob!, a multi-user mobile and locative (i.e. pervasive) game. I also give an overview of the state of the art in the world of pervasive gaming. I introduce the types of pervasive games, the technologies are built upon and the positioning techniques. I finish by mentioning the challenges such games face, that is dealing uncertainty, privacy issues and making the games and technology engaging.

Previously I wrote a post-mortem on CatchBob! entitled Building a mobile, locative, and collaborative application. It describes the whole development process, from the technical architecture to the user perception of the game. it addresses a large audience. Both techies and non-tech-savvy people should not be afraid to have a glance at it.

CatchBob!

CatchBob!, a collaborative mobile and locative game, was developed at the CRAFT, a learning technology lab, with a focus on technologies for collaboration, at the Swiss Federal Institute of Technology in Lausanne. It was set up to test how a location awareness tool modifies the group interactions and communications, the way they perform a joint task as well as how they rely on this spatial information to coordinate. Series of experiments are run on this platform so that we can figure out how location awareness impacts group collaboration.


CatchBob! players firing up their PocketPCs and planning a strategy to start the game (figure 1). CatchBob! on TabletPC, a player annotating the map (figure 2), CatchBob! interface on TabletPC (figure 3). An animation based on screenshots of the interface with no awareness tool (figure 4). Players are self-confronted with a replay of their paths on the campus taken from the server log (figure 5).

Pervasive Gaming

Pervasive gaming was first the vision of Swedish company It's Alive!, meaning location-based games that surround you, 24 hours a day, everywhere. When you walk down the street, you're walking through an adventure world draped on top of the real world, and people you meet may be characters in the same game you're playing. Pervasive games are built upon three core technologies: mobile devices, wireless communication, and sensing technologies that capture players’ contexts. It is actually the blend of technologies combined with the location-based and often public nature of game play, gives pervasive games their distinctive identity [Bridging the Physical and Digital in Pervasive Gaming].

Types of Pervasive Games

The academics in the locative mobile games have been experimenting their games for a couple of years now. Major achievements include Uncle Roy All Around You, and Can You See Me Now? and Human Pacman. A regularly updated list of location-based games is available on IN-Duce.

Pervasive game pioneer Steve Benford, categories the various forms of pervasive games:


Uncle Roy All Around You (figure right) is a touring artistic type of pervasive game. Street Players use handheld computers to search for Uncle Roy, using the map and incoming messages to move through the city. Human Pacman (central figure) is a reality augmented pervasive game mapped from an old arcade game. Savannah (right figure) is an educational game in which children play at being lions in a savannah, navigating the augmented environments with a mobile handheld device.

The Technologies

In Future Location-Based Experiences, Steve Benford specifies the three core technologies that location-based experiences build on : mobile devices, wireless networking and location sensing.

Mobile devices vary greatly in terms of computing power, display size, form factor, portability and battery life. The displays make digital content available to players as they move through the physical world. CatchBob! uses PocketPCs and TabletPCs. Cell phone is the most ubiquitous computing platform and therefore is an excellent candidate for pervasive games deployment.

In wireless networking three broad classes of technology seem likely to dominate the field in the immediate future: cellular telephony, Wireless Ethernet and Bluetooth. CatchBob! is based on 802.11b technology.

There is a wide variety of location sensing technologies that mainly include GPS, Wireless network, Ultrasonic systems, RFID tags, accelerometers, and vision techniques. The problems of using radio frequency signals-based location, because it is hampered by inherent technology problems such as limits on coverage, signal interference, and reliance on infrastructure, and by broader issues such as privacy concerns.

Positioning

The obvious solution for positioning is to use GPS (Global Positioning System). However it requires a line of sight to the sky therefore does not work indoors nor in dense urban environments. Using GPS was out of question for CatchBob!.

Most current location systems do not work where people spend much of their time. GPS does not work indoors and works poorly in many cities where the so called “urban canyons” form by building prevent GPS units from seeing enough satellites to get a position lock. Despite the name, GPS is less ubiquitous than mobile phone technologies in places of human significance. Intel Research's Place Lab initiative promotes a mix approach to positioning to reach ubiquitous availability of location information. By 2008 the European Union will deploy Galileo, a next-generation GPS system that promises greater accuracy and operation covering both indoors and out, due to stronger radio signals that should penetrate most buildings.

There are basically four ways to do indoor positioning: with infrared beacons, radio beacons, ultrasound systems and video-based systems. GPS can be combined or augmented with these technologies to extend the coverage or for calibration.

Pre-setting, configuring and calibration of a locative game can easily become a challenge. For CatchBob! we avoided using extra infrastructure for not spending continuous amount of time on these tasks. Therefore, Radio beacon based positioning with Wi-Fi was the only alternative.

This table below shows the results of accuracy and coverage test for radio beacon (Wi-Fi and GSM) positioning using Place Lab without any special installation and calibration in 3 types neighborhoods. Taken from Place Lab: Device Positioning Using Radio Beacons in the Wild.
802.11
GSM
accuracy
coverage
accuracy
coverage
Urban
20.5 m
100%
107.2 m
100%
Residential
13.5 m
90%
161.4 m
100%
Suburban
22.6 m
42%
216.2 m
100%

In CatchBob! we question if sharp positioning accuracy improves the collective performance on the task. Chris Heathcote (in his talk 35 ways to find your location) advocates for the use of technology for quick rough location and then the use of people to refine. How accurate the positioning should be to be useful, especially when humans can read/recognize 50-100m? He says that instead of throwing technology at the problem we should match the needs to methods for the developer and the user (appreciate the toolbox). He mentions the different measures of location: accuracy, availability, reliability/trust, output useful to humans, output useful to computers, acquire or refine, and ask us to reflect on each of this category. What is good enough, useful and useable by the players?

Techniques

There are two main solutions used in radio beacon positioning. The first one, which is called the empirical model, is based on storing pre-recorded measurements in a database. Before any location positioning can be performed, a radio map is created. This map stores the signal strengths at each particular location from all access points that can be read from that area into a database. When a device requests a location, it matches signal strengths from all the access points that it can read with the database.

The second solution is called the propagation model. It is based on the degradation of the signal strength of a radio wave over distance in space. The accuracy of this model is dependent on the algorithm that is used.

CatchBob! is based on the propagation model. Mainly because the campus network we use changes and expends regularly. Constantly updating and calibrating the radio map would have been too fastidious. I used a simple triangulation to compute the positions of the players. Nowadays, triangulation is used for any kind of geometric approaches for location. I only implemented a very simple Centroid algorithm. I position the user at the center of the scanned nearby access points by computing an average of their x, y location. In addition I weighted by the received signal strength during a scan. Accuracy Characterization for Metropolitan-scale Wi-Fi Localization talks about positioning algorithms including also Fingerprinting and Particle Filters. A rough 2D positioning accuracy of 5-30 meters was enough to achieve the collaborative task of CatchBob!.

Ekahau, a Wi-Fi positioning company claims to reach an accuracy of 1-2 meters. However, this precision is achieved on tightly controlled conditions, with many hours of installation and calibration.

3D positioning is hard to achieve with a propagation model. The problem is that the third dimension, the altitude component of location, is a discrete data. The current trend is to generate“2.5” dimension estimates in which altitude is represented with a symbolic name such as “Parking Level A” or “3rd floor” that are more meaningful than a coordinate-based altitude like 34.6 meters above sea level. Since in CatchBob! my positioning algorithm was already giving approximate 2D estimates, I did not want to add another factor of approximation. I doubt the players would have understood the 3D positioning failure, especially when we ask them to trust an already approximate positioning system.

Perception of Uncertainty

Any network programming is about managing availability, latency and reliability. On top of that, pervasive application developers have to learn how to deal with uncertainty of location and connection. The have to work around or play with a limited coverage, accuracy, variation of performance over both space and time.

There was a lot to learn listening the CatchBob! participants talking about the application and the interface. There was a gap between the player ’s expectations and what the tool really did. Sometimes it lead to the lost of trust on the tool “it did not act like I thought it would have” and frustration “sometimes it refused to work”. An unreliable network is sometimes not perceived as a disturbance but as a failure. A couple of users even blamed the tool for the lost of connectivity. It is difficult to make understand the state of connectivity. Mainly because it is hard to detect a break of connection. A client can still detect nearby Wi-Fi access points, while not having enough signal to send and receive data. In that case, displaying the signal strength might not be enough.

The positioning accuracy was not always understood and disturbed some players “I did not move physically, but I moved on the map” and “The proximity to Bob changed even though I did not move”. This was rather a surprise. Players came with a pre-conception on the quality of positioning systems. It is still unclear to me why. GPS systems have an accuracy of 5-10 meters. Neither is it clear how bad and good positioning accuracy could be displayed.

One trend in location-based mobile games is to play with the tolerance and reliability of the technologies used. Positioning errors, maps approximations and network unreliability become part of the games. Matthew Chalmers' team at the University of Glasgow came up with the concept of seamfulness. A seamful game is a GPS and Wi-Fi based game that harness negative aspects of infrastructure technologies, which are normally concealed and unexplained, and present them as game features allowing users to explore and understand them.

In Bill (figure right), players must develop an understanding of the network coverage and the effect of signal strength in order to successfully play the game.

Privacy

Among with accuracy and cost, privacy, is a major concern in phone-based location systems. There is a high demand from users for security, privacy and trustworthiness. There are three main privacy issues encountered by pervasive system [Privacy and Security in the Location-enhanced World Wide Web]:

Concerns on privacy often come from the lack of control and proper feedback about "what is controlling what, what is connected to what, where information is flowing, how tis is being used". It is possible that experience breeds higher level of trust. Like it happened for the adoption of the Internet, novices were often very concerned about privacy, but after a year of experience online, they showed an increase in the number of trusting activities performed online.

Techniques to address end-user concerns about location privacy are not trivial. The positioning technique used for my locative application such as CatchBob! matches the privacy model of GPS. The devices compute location estimates autonomously rather than divulging their location to an infrastructure. Users know when they divulge their location. Basically players must be able to trust that their location is not disclosed without their explicit consent and it should avoid reliance on third parties who my not be perceived by players as thrustworhy.

Engaging games and technologies

Players mentioned in their post-interview that they genuinely enjoyed CatchBob! The interface and technology are fun to explore, but the game did not provide any real engaging scenario. In video games there is a strong sense of challenge and a balance between the individual challenge and a team challenge. Like in real life, we tend to collaborate to achieve our own personal goals.

The game scenario put aside, would the technology we used and the way we interfaced it be embraced by a majority of people? This question is the key to any user-centric technology, but even more especially so in the “everyware” world, because ubiquitous, pervasive and geolocated systems are about to enter into people’s lives. They will tend to become inevitable, but it does not mean they will be accepted. The danger is that the pervasive technologies present users with unacceptable difficulty, confusion and uncertainty.

A problem in the mobile Internet is that it is not open and user-driven. It is a shame because it prevents users to actively shape media the media. They are passively consuming what is provided by a few. Quoting Howard Reingold in Mobile and Open: a Manifesto: "If hardware can't be hacked and software is locked away from individuals by technology or law, users won't be free to invent".

Conclusion

Building a pervasive game is a challenging engineering task because it touches many aspects of software computing and game design, from networking to the fields of human-computer interaction, and social computing like carrying the sense of connectivity or providing a social context.

Current technologies have their imperfections and we have to play around them or mix them. The limits are now known. Engineers have been doing their work by providing frameworks and toolkit for Wi-Fi positioning and for data communication. Building location-based application is being simplified. CatchBob! is a clear example of these achievements. Nevertheless, I lot still needs to be done for locative games to reach maturity, to strengthen the link between the numerical and physical world. No matter how good the technology is, it has to be enjoyable and useable, it has to be engaging!

Fabien (dot) Girardin (at) epfl (dot) ch