Transposition of Location-based Games: Using Procedural Content Generation to deploy balanced game maps to multiple locations

Highlights

• Using PCG to tackle the challenge of transposing maps of LBGs while focusing on maintaining game balance.

• Formulation of the Map Transposition challenge as a Graph Matching Problem and the use of a Genetic Algorithm to tackle it.

• Experiments to validate the proposed approach with four distinct games and several subjects.

Abstract

Location-Based Games (LBGs) rely on the player’s location to change its game state. Developing worldwide LBGs is a challenging task due to the need to deploy game instances in multiple locations, while maintaining the same game balancing, features, and even correlations between locations of the game and the real world. Hence, it is virtually impossible to manually design interactions, challenges, and game scenarios for every place a player is at. As a result, even modern LBGs still present huge balancing differences between regions and avoid exploring the competition between players like other game genres. This work uses Procedural Content Generation to transpose maps of LBGs while focusing on maintaining their game balancing. We convert LBGs into a game model based on a directed weighted graph using information about Points-of-Interest (POIs) around the players’ location. This game model is simplistic and lightweight, which makes it suitable to wearable and IoT devices that are becoming popular among LBGs. Moreover, we use a Genetic Algorithm to generate a corresponding LBG instance with similar game balancing. To validate the proposed approach, we designed four LBGs with distinct features, gameplay, and mechanics, and conducted an experiment that required samples to compare maps generated by the algorithm in different locations. Results indicate that games with similar game balancing score higher and that the number of POIs has a significant impact on the performance of the approach. Finally, we can conclude that this work contributes to improving the development of LBGs by helping to mitigate the challenge of transposing LBGs while maintaining game balancing.

Introduction

Since the release of the first iPhone in 2007, the smartphone has become a technological hub that includes several sensors, considerable processing power, and is capable of countless applications. In a short time, many game companies realized that smartphones and tablets could become a platform for playing games. Previously, games were played in specific places using devices such as keyboard, mouse, joystick, PCs, and consoles [1]. Nowadays, smartphones have enabled games to be played at any time and everywhere. Consequently, many games have been ported or developed specifically to the mobile market. In fact, smartphones have impacted the way people play games and have boosted the game industry due to its immense potential as a gaming platform. Additionally, the increase in processing power and the popularization of sensors in modern smartphones has allowed the implementation of new game genres, such as Pervasive Games.¹

Taking advantage of the resounding success of smartphones, the Internet of Things (IoT) has made significant strides in recent years. Gubbi et al. [3] define IoT as the interconnection of sensors and actuators, providing the ability to share information across platforms, acting as a unified framework. IoT is primarily motivated by the growth in the number of sensor-based smartphones, and also from the spread of wearable devices (e.g., smartwatches), physical sensors (e.g., sensor tags, Bluetooth beacons) and smart objects (e.g., Amazon Alexa, Philips Hue lights). Recently, the game industry started to use IoT devices to increase pervasive game experience, for instance, incorporating sensors in tabletop games or connecting wearable devices (e.g., smartbands) to smartphone games.

This paper focuses on a prevalent subtype of Pervasive Games called Location-based Games (LBGs), which we define in this work as games that use players’ location to modify the game state during runtime. As a result, players have to physically move to progress in the game, thus establishing a link between virtual and real worlds. Although LBGs are games, they also have characteristics and challenges of classical LBS (Location Based Services). For example, an LBG must provide system adaptation according to the user’s context, navigation on world-scale maps, and real-time updating of a large mass of sensed data (i.e., the position of users and Non-player characters — NPCs) [4].

Despite the release of more LBGs in the last two years (e.g., Harry Potter: Wizards Unite, Jurassic World Alive, etc.), they still represent just a fraction of the mobile gaming industry. In fact, only a few game studios are currently venturing into this area. Nevertheless, cost and time are key problems to produce digital games, with many titles taking more than 3 years of development, and costs that exceed tens of millions of dollars [5], [6]. This scenario is not different for LBGs, with Niantic and Google spending roughly $30 million dollars in the development of Pokémon GO [7]. As a result, game studios are continuously trying to reduce expenses to both minimize risks and increase profits.

In addition, producing an LBG is even more challenging as it presents features and issues that are absent in most games (e.g., LBS features). For instance, developers have to create maps available in a vast number of places, cope with tracking issues, and adjust game balancing in different regions. Furthermore, new wearables, such as the Pokémon GO Plus Strap, and IoT are expected to enhance monitoring and gameplay in LBGs. Using new sensors and IoT in LBGs has the potential to increase the players’ immersion and provide data that are key to refine the Location-based Services on these games [8].

In general, an LBG can be played virtually everywhere, provided their game elements are correctly mapped to the proper locations in the real world. To address this challenge, a strategy employed by some games is to use a database containing Points-of-Interest (POIs) scattered around the globe. The main drawbacks of this approach are the complexity, time, and cost to build and maintain such a large database of POIs. However, a low-cost and more feasible solution could involve Procedural Content Generation (PCG), a field of Computer Science that proposes methods and algorithms used in the creation of content automatically, thus reducing costs and time of production.

Besides, maps of LBGs must fulfill requirements that suit aspects of both real and virtual environments, such as avoiding private properties and unreachable areas, and adapting to varying interactions implemented by the games. Furthermore, developers must consider the time and effort to physically move between sites as a crucial factor to game balancing.²

Consequently, there are two main challenges that hinder the mass production of new LBGs. First, the need to map the game to countless places while preserving their game balancing; and second, the high costs and time required to develop worldwide LBGs [11]. We believe that these challenges have impaired many game studios and independent developers from creating more games. This scenario is the key motivation for this research, and an alternative to alleviate this problem called Transposition of maps for LBGs is explored. It uses PCG to create maps of LBGs according to the location of each player. In this case, we propose a general game model that works as a generic representation for distinct LBGs and includes only the data required for the transposition, thus being suitable to wearable and IoT devices. Our game model is defined as a weighted graph, and the transposition challenge is modeled as a particular case of Graph Matching Problem that can be solved using optimization algorithms.

In summary, the main contributions of this work are as follows:

• Using PCG to address the challenge of transposing maps of LBGs while focusing on maintaining game balance;

• The use of a Genetic Algorithm to tackle the Map Transposition challenge as a Graph Matching Problem;

• Experiments to validate the proposed approach with four distinct games and several subjects.

The paper is organized as follows. Section 2 presents works that are the basis for the development of this research. Section 3 introduces the proposed method, while Section 4 formulates the problem, and details the use of Genetic Algorithms to tackle the challenge. Section 5 shows evaluations conducted with varying input sizes and several users to validate the proposed method. Lastly, Section 6 concludes the paper and points to future works.