Adapting Playware to Multiple Players

Recent Researches in Communications, Electrical & Computer Engineering Adapting Playware to Multiple Players ARNAR TUMI THORSTEINSSON Center for Play...
Author: Norah Walker
1 downloads 0 Views 1MB Size
Recent Researches in Communications, Electrical & Computer Engineering

Adapting Playware to Multiple Players ARNAR TUMI THORSTEINSSON Center for Playware Technical University of Denmark Building 325, 2800 Kgs. Lyngby DENMARK [email protected]

HENRIK HAUTOP LUND Center for Playware Technical University of Denmark Building 325, 2800 Kgs. Lyngby DENMARK [email protected] http://www.playware.dk

Abstract: - With the creation of playware as intelligent hardware and software that creates play, it is possible to adapt the play tool to the individual user, and even to multiple users playing at the same time with the play tool. In this paper, we show how it is possible to implement adaptivity in modular playware, and allow the playware to adapt to the user´s level of competency in multi-player games. The games are physically interactive games where users may have different levels of competencies due to different physical abilities e.g. between age groups and genders. Indeed, the work gives evidence to such differences, and argues that adaptivity is needed to make games fit to the individual users in both single-player games and multi-player games. As a case study, we implemented such adaptivity on modular interactive tiles for the single-user game ColorTimer, and for the multiple-user games PingPong, in which two players are competing against each other. In the game, the speed will change to fit the level of the individual player, so that the game may be faster for one player, and slower for the opponent player to match the level of the player, so that each player is challenged at or just above the players’ level of competency. Key-Words: Adaptivity, playware, games, modularity

1 Introduction

2 Modular Interactive Tiles

People may interact in different ways with the objects and environment that surrounds us. The differences in interaction style and pattern may have different reasons such as cultural differences, physiological differences, cognitive differences, and differences in experiences amongst people. Due to the differences in interaction style amongst people, it seems appropriate that the objects which surround us should be able to change themselves to fit the interaction style of the individual user of these objects. In order to make objects change themselves to fit the interaction style of the user, it necessary to understand the differences in the way that people interact, and what differences may be the basis for making adaptation. Based on such knowledge, it is further important to investigate engineering methods for making adaptation of objects to the individual user and confront the challenge of how to make adaptation when more users are interacting with an object. In the following, we will show how modular playware may be the basis for creating adaptation to multiple users’ interaction at the same time with the playware object. With simple playware games, we show that the adaptation will speed the game up and down to find the appropriate level that matches the reaction speed of the individual player. The appropriate level will change with game/interaction complexity, and adaptation will automatically find the appropriate level for the individual player, even in multiplayer games.

In order to investigate adaptive modular playware, we developed the modular interactive tiles [1, 2]. The modular interactive tiles can attach to each other to form the overall system. The tiles are designed to be flexible and in a motivating way to provide immediate feedback based on the users’ physical interaction, following design principles for modular playware [3, 4].

ISBN: 978-960-474-286-8

Fig. 1. Modular tiles used for feet or hands gaming interaction.

Each modular interactive tile has a quadratic shape measuring 300mm*300mm*33mm – see Fig. 1. It is moulded in polyurethane. In the center, there is a quadratic dent of width 200mm which has a raised circular platform of diameter 63mm in the centre. The dent can contain the printed circuit board (PCB) and the electronic components mounted on the PCB, including an ATmega 1280 as the main processor in each tile. At the center of each of the four sides of the quadratic shape, there is a small tube of 16mm diameter through which infra-red (IR) signals can be emitted and received (from neighboring tiles). On the back of a tile there are four small magnets. The magnets on the

266

Recent Researches in Communications, Electrical & Computer Engineering

least one master tile if communication is needed e.g. to game selector box or a PC. With these specifications, a system composed of modular interactive tiles is a fully distributed system, where each tile contain processing (ATmega 1280), own energy source (LiIo polymer battery), sensors (FSR sensor and 2-axis accelerometer), effectors (8 colour LEDs), and communication (IR transceivers, and possibly XBee radio chip). In this respect, each tile is self-contained and can run autonomously. The overall behavior of the system composed of such individual tiles is however a result of the assembly and coordination of all the tiles.

back provide opportunity for a tile to be mounted on a magnetic surface (e.g. wall). Each side of a tile is made as a jigsaw puzzle pattern to provide opportunities for the tiles to attach to each other. The jigsaw puzzle pattern ensure that when two tiles are put together they will become aligned, which is important for ensuring that the tubes on the two tiles for IR communication are aligned. On one side of the tile, there is also a small hole for a charging plug (used for connecting a battery charger), including an on/off switch. There is a small groove on the top of the wall of the quadratic dent, so a cover can be mounted on top of the dent. The cover is made from two transparent satinice plates on top of each other, with a sticker in between as visual cover for the PCB. A force sensitive resistor (FSR) is mounted as a sensor on the center of the raised platform underneath the cover. This allows analogue measurement on the force exerted on the top of the cover. On the PCB, a 2 axis accelerometer (5G) is mounted, e.g. to detect horizontal or vertical placement of the tile. Eight RGB light emitting diodes (LED SMD 1206) are mounted with equal spacing in between each other on a circle on the PCB, so they can light up underneath the transparent satinice circle.

Fig. 3. Assembly of the modular interactive tiles as a simple jigsaw puzzle.

The modular interactive tiles can easily be set up on the floor or wall within one minute. The modular interactive tiles can simply attach to each other as a jigsaw puzzle, and there are no wires. The modular interactive tiles can register whether they are placed horizontally or vertically, and by themselves make the software games behave accordingly. Also, the modular interactive tiles can be put together in groups (i.e. tiles islands), and the groups of tiles may communicate with each other wireless (radio). For instance, a game may be running distributed on a group of tiles on the floor and a group of tiles on the wall, demanding the user to interact physically with both the floor and the wall.

Fig. 2. PCB and components of a modular interactive tile.

The modular interactive tiles are individually battery powered and rechargeable. There is a Li-Io polymer battery (rechargeable battery) on top of the PCB. A fully charged modular interactive tile can run continuously for approximately 30 hours and takes 3 hours to recharge. The battery status of each of the individual tiles can be seen when switching on each tile and is indicated by white lights. When all eight lights appear the battery is fully charged and when only one white light is lit, the tile needs to be recharged. This is done by turning of the tiles and plugging the intelligent charger into the DC plug next to the on/off switch to recharge each tile. On the PCB, there are connectors to mount an XBee radio communication add-on PCB, including the MaxStream XBee radio communication chip. Hence, there are two types of tiles, those with a radio communication chip (master tiles) and those without (slave tiles). The master tile may communicate with the game selector box and initiates the games on the built platform. Every platform has to have at

ISBN: 978-960-474-286-8

3 Adaptation We propose three different game scenarios with adaptation of games at run-time (see Figure 4):

Figure 4. Three different kinds of run-time adaptation for single player game, cooperative game and competitive game.

• Single Player Games (Serial Adaptation): When a game is played by one player at a time, it can possibly be beneficial if the game will adapt itself to the speed of that player, instead of the player having to adapt to the speed of the game. Such adaptation is important because

267

Recent Researches in Communications, Electrical & Computer Engineering

many physical games may be used by people with decreased motor skills and/or other disabilities, as well as by healthy individuals. Since those disabilities may limit some players more than others, the same game would not fit all players well without some kind of adaptation.

In ColorTimer, a tile will shine up and the user has to hit it. With every fast hit of a lit tile by the user, the program will advance to the next level and will speed up. If the user misses or hits slowly, the game will go down a level again and slow down again. The game is implemented so that it can be played with 1, 2, 3 or 4 tiles lighting up at the same time (i.e. the game can be more or less complex).

• Cooperative Games (Parallel Adaptation): When two players play a cooperative game together, the game may tend to become too difficult for the slower player. If a game could distinguish between these two players and adapt the game to their speed individually, both players could work together, on their own speed. This same kind of adaptation can possibly be used for individuals going through rehabilitation after a stroke. Stroke victims often lose some or all mobility in one half of their body. Here, we can describe the less mobile part of the person’s body as the slow player, while the healthy part of the body is the fast player. In that case the game could adapt to each part individually, allowing this person to practice both parts of his/her body simultaneously based on the different motor skills of the two parts of the body.

Figure 5. The ColorTimer game running with three lit tiles at different times.

There are no limits on the amount of tiles able to be connected together for this game, but for testing purpose the testing platform was set to 8 tiles, arranged in a pattern of 2x4 tiles.

• Competitive Games (Biasing): When two players of very different skill level play against each other it is an open question how to set the difficult level of the game. If the game is tuned to meet the requirements of the slower player, the faster one will soon feel bored since the game does not push his abilities. However, if the game is tuned to meet the requirements of the faster player, the slower one will not be able to keep up and will thus soon lose interest in the game itself. If a game could distinguish between those two players and bias the game in such a way that both users would be challenged, it could prolong the playing as it would become more entertaining for both players.

PingPong: We implemented a similar type of speed adaptation used in the ColorTimer game into a more structured two player game, namely PingPong. One of the first computer games ever made, Ping-Pong, was used as a reference for the PingPong. In the original game, the two players each control a small pad on the screen. A ball bounces back and forth and the players are supposed to defend their goal by adjusting the paddle so the ball will hit it and bounce back. In PingPong, we included game elements, e.g. using lives as points, and created more direct competition between users by structuring the game as a two player game. The program was to be able to keep both players equally challenged during the game by biasing it individually for each player. In the game, each player stands at one end of the shorter sides of rectangular tile cluster on the floor. The row in front of them is the defense line. One tile would light up at a time and bounce from one end to the other. As the light hits the defense line, the user is supposed to step on it as quickly as possible, causing the ”ball” to bounce in the opposite direction. Each player has eight lives, displayed by the amounts of lit LEDs on the defense line between rounds. For this game, the adaptation is in the form of biasing. When a defense tile lights up, a timer is started. As the user hits the tile, the timer is compared to the current player time. If the timer has counted less than 20% of the current

In order to make investigations on the role of adaptation, the adaptation was kept simple focusing on the speed of the user, incorporating adaptation of colors. Also, pattern complexity was investigated to see if it played a role for how well the players reacted to the games. Other possibilities could be to adapt e.g. the distance between light activated tiles or the pressure needed to activate the tiles, but in the work presented here, the main focus was on the adaptation of the speed and changes in pattern complexity in the games, which were designed for these adaptation experiments.

3.1

Game Design We developed some simple games, which would be easily scalable and allow us to make the adaptation tests. The games included ColorTimer (used to test serial adaptation and parallel adaptation) and PingPong (used to test biasing).

ISBN: 978-960-474-286-8

268

Recent Researches in Communications, Electrical & Computer Engineering

lit time of the tile, the time of the player is reduced by the same amount, making the game faster. If the timer has counted more than 80% of the current lit time of the tile, the time is increased making the game slower. Figure 6 shows three stages of the PingPong game. The first one shows the startup where the game selects the shorter sides as defense sides, shown as red and green. The second one shows the ”ball” on the defense line of one of the players. The third one shows how the field displays lives left. Notice how the color of the lives is slowly changing as they get fewer.

(countdown time on y-axes) over time (x-axes). The graphs indicate the optimal speed of the player for each game, and thereby indicate the difference in reaction speed between games. The graphs show a clear difference between the children and the adults, with adults being much faster than the children (the children are approximately half as fast as the adults). This may indicate that the children´s reflexes / sensory-motor coordination skills are not fully developed. Another interesting thing is the adult males are in all cases considerably faster than the adult females. This is not always the case for the children, since in some cases the boys are faster and in others the girls are faster.

Figure 6. Three stages of the PingPong game.

4 Tests The research questions related to adaptivity were tested on two smaller target groups, each group with both males and females and each group at a different age. The first test was done using 11 adults, six females and five males, age 20-30, and six children, three boys and three girls, all 6-7 years old from a school class at Nyboder Skole in Copenhagen. Each of them was asked to play the ColorTimer game three times using their using their feet. Each game had different number of lit tiles, from one to three. The player was supposed to hit the lit tiles as fast as possible. Each hit resulted in the tile turning off and another random one lighting up instead. The length of each game was selected to be no longer than 120 hits, but since it was important to get at least 10 hits per tile for comparison, the random generator was biased so that each tile would not light up more than 15 times. The length of the game was decided from a couple of trial runs. With more than 120 hits, the player would become too tired, and any less would not fully guarantee enough points to evaluate from. The participants were allowed as much time as they needed between games to catch their breath, in order to reduce changes caused by the players fatigue. The data of the experiments was collected in a graph of comparison, Figure 7. Here, the children are colored by gender and compared to adult data, also colored by gender. The left graph shows the average of all players, and the age and gender average is shown on the right graph. The graphs show how adaptation is working in speeding the game up

ISBN: 978-960-474-286-8

Figure 7. Average countdown time found by the adaptation process when users are playing with their feet (left: all individual players, right: average sorted by age and gender).

These comparative graphs on figure 7 clearly show how adaptation is working automatically finding and adjusting to different, individual levels for the different user groups (men, women, boys, girls). We can only use this as an indication, but it would be interesting to see if this pattern holds for a larger test group. The PingPong program was tested using a group of six adults, four males and two females playing against each other. For testing purpose, each game was four rounds long.

269

Recent Researches in Communications, Electrical & Computer Engineering

The participants could decide who played together, so the results include games played between two male users as well as games played between a male and a female.

red user. This indicates the adaptation is working as expected. This data indicates not only that it is possible to create adaptive games using modular interactive tiles, but by doing so one may possibly be able to keep the users more challenged by tuning the game to suit the individual. Figure 9 shows the other four games that were played. Each game is four rounds long, but the games may have different lengths. The game keeps biasing the players, usually speeding up the game towards one of the players. Usually these speedups occur early on in each round, indicating that the initial startup speed is too slow for some of the players. Another interesting point is when the game adapts (either slows down or speeds up), it sometimes readapts very soon, hitting a very similar speed to where it was before adapting in the first place. This can be seen for example in the first graph around the hit 34 and hit 76 for the red player. This is interesting because it means that even though the user gets in a quicker hit than his average speed, he may not be able to keep that speed up, causing the game to readjust to the previous speed. Another interesting point is that for this more complex game with the feet, there is no apparent difference in the performance between the genders.

Figure 8. Adaptation of game speed for two individual users in the competitive game, PingPong.

Figure 8 shows a typical game of four rounds. The speed of the game was reset to 500ms after each round of the game. The adaptation in this competitive game is in the form of biasing the game. The game listens for a hit when a tile in the defence-line is lit and compares the hit time to the current time each tile is lit for this player. If the hit occurs in less than 20% of this player’s time, that time is reduced proportionally to the hit-time. If the hit occurs after more than 80% of the player’s time, the time is extended proportionally to the hit-time. As can be seen from the figure, the game becomes biased, and the biasing seems consistent. For instance, the blue user who seems to be an average user, coping well with the constant speed increase, but the player not very often fast enough or slow enough for the game to have to step in and bias the game for this user. It is only in the fourth round the blue user’s response seems to become faster, causing the game to speed up quickly. The red user on the other hand is quite a lot faster than the blue user. For all four rounds the game biases against that user heavily, causing the time of each light to become at least 100 ms shorter than towards the blue player, possibly causing the game to become more interesting for that player since he is challenged based on his own speed, and is not limited to the speed of the opponent. When analyzing the data of such a game more closely, we can see that the two fast responses by the blue user (at hits 1 and 23) result in a steeper change in the speed, and that for the red user, the good hits (at 4 and 6) result in a very steep change in speed, and the faster the response, the steeper the change. However, when (at hit 28) the red user is slower than the higher margin, the game slows down again for the

ISBN: 978-960-474-286-8

Figure 9. Adaptation of game speed for four player pairs playing the competitive game, PingPong.

270

Recent Researches in Communications, Electrical & Computer Engineering

5 Conclusion The results indicate some interesting differences between age groups and gender groups. The fact that males tend to play faster than females in the ColorTimer games tested here can have different explanations and we must remember that the sample is very small, and therefore the results can only be indicative. A simple explanation may be that this particular game in its play nature favours males, who may possibly respond with a more competitive attitude to this game, and therefore are more motivated to perform fast. Another explanation may be that there are physiological differences in speed and strength between the gender groups in the small sample set. The difference in performance between age groups (children and adult) seems clearer and is also expected due to differences in physical and cognitive skills between the two age groups. Despite the small test set, the results are important as a proof of existence of differences and of the need for adaptation. The fact that there are individual differences (e.g. down to differences between as few as two individuals) makes the results significant for the development of games and interaction. It clearly indicates that it is necessary to adapt the game and interaction, if we desire to make the most appropriate game and interaction for the individual. Indeed, with the work, we have tried to show how playware may adapt to an appropriate level for the individual user. The results show that the adaptation will speed the game up and down to find the appropriate level that matches the reaction speed of the individual player. The appropriate level will change with game/interaction complexity, and adaptation will automatically find the appropriate level for the individual player. The work presented here shows by a simple evidence of existence that we may develop playware equipment that by itself in an automatic, dynamic way makes the appropriate zone of proximal development [5, 6] for the individual human being. It adjusts itself so that it can possibly become interesting, challenging, and motivating for the individual user, because it gives the right playful challenge for that particular individual. Even when two players are playing together or against each other, we have shown indications of how adaptation may work to adapt the game to the two individual levels appropriate for each of the players, who are playing the same game concurrently on the same platform.

References: [1] H. H. Lund. “Modular Robotics for Playful Physiotherapy,” in Proceedings of IEEE International Conference on Rehabilitation Robotics, IEEE Press, 2009, 571-575. [2] H. H. Lund, M. D. Pedersen, and R. Beck. “Modular Robotic Tiles – Experiments for Children with Autism”. In Proceedings of 13th International Symposium on Artificial Life and Robotics (AROB'13), ISAROB, Oita, 2008. [3] H. H. Lund, T. Klitbo, and C. Jessen. “Playware Technology for Physically Activating Play”, Artificial Life and Robotics Journal, 9:4, 165-174, 2005 [4] H. H. Lund, and P. Marti. “Designing Modular Robotic Playware.” In Proc. of 18th IEEE International Symposium on Robot and Human Interactive Communication (Ro-Man 2009), IEEE Press, 2009, 115-121. [5] L.S. Vygotsky. Mind in Society: The development of higher mental processes, Harvard University Press, Cambridge, MA, 1930-35 (1978). [6] L. S. Vygotsky. Thought and language. Cambridge; MIT Press, 1986.

Acknowledgement The authors would like to thank colleagues at the Center for Playware, DTU Electrical Engineering for interesting and valuable discussions.

ISBN: 978-960-474-286-8

271