HybridPLAY: A New Technology to Foster Outdoors Physical Activity, Verbal Communication and Teamwork

This paper presents HybridPLAY, a novel technology composed of a sensor and mobile-based video games that transforms urban playgrounds into game scenarios. With this technology we aim to stimulate physical activity and playful learning by creating an entertaining environment in which users can actively participate and collaborate. HybridPLAY is different from other existing technologies that enhance playgrounds, as it is not integrated in them but can be attached to the different elements of the playgrounds, making its use more ubiquitous (i.e., not restricted to the playgrounds). HybridPLAY was born in 2007 as an artistic concept, and evolved after different phases of research and testing by almost 2000 users around the world (in workshops, artistic events, conferences, etc.). Here, we present the temporal evolution of HybridPLAY with the different versions of the sensors and the video games, and a detailed technical description of the sensors and the way interactions are produced. We also present the outcomes after the evaluation by users at different events and workshops. We believe that HybridPLAY has great potential to contribute to increased physical activity in kids, and also to improve the learning process and monitoring at school centres by letting users create the content of the apps, leading to new narratives and fostering creativity.


Introduction
Game-based technologies for entertainment and/or edutainment are not only for adults, but a great market exists centered on the early ages. A recent study [1] reveals that 38% of children under the age of 2 have used a mobile device for playing games, watching videos or other media-related purposes. By the age of 8, 72% of children have used a smartphone, tablet or similar device. By contrast, the use of playgrounds, ballparks, etc. decreases from the age of 6, which contributes to reduced physical activity and increased childhood overweight/obesity and creates a rift with children's social environment.
From the social point of view, games are proven learning mechanisms-fundamental at early ages-to acquire skills and abilities such as communication and social interaction, and especially to be temporarily transformed into "other" people and acquire the ability to understand the view of the "other", and thus substantially enhance our ability to understand the world and how to access it [2]. Although these benefits are evident in traditional games, in video games the problem is that their prolonged use can potentially lead to develop some addictions, especially in the so-called massive multiplayer online role-playing games, with the consequent risk of isolation and inability of children to develop social and communicative relations in the physical space; especially when children and While designing the sensorial elements of HybridPLAY, the following five key points have been taken into account: 1. Versatile placement: the sensors can be easily placed on any playground element without instructions, taking into account the diversity of materials, current condition, surface finishes, dimensions, and positions. 2. Intuitive single solution grip: HybridPLAY is mainly for kids, so it requires an intuitive grip, allowing children to easily change the game whenever they want with minimal installation time. 3. Customizable and suitable for all ages: though initially intended for children, HybridPLAY is versatile and has an immense range of potential users. 4. Compact and weather-resistant: naturality and impulsiveness are key in our games, so there's no time for accessories. 5. Simple Design from a single mould: a simple design lets us keep costs down, and makes the project viable. While designing the sensorial elements of HybridPLAY, the following five key points have been taken into account:

1.
Versatile placement: the sensors can be easily placed on any playground element without instructions, taking into account the diversity of materials, current condition, surface finishes, dimensions, and positions.

2.
Intuitive single solution grip: HybridPLAY is mainly for kids, so it requires an intuitive grip, allowing children to easily change the game whenever they want with minimal installation time.

3.
Customizable and suitable for all ages: though initially intended for children, HybridPLAY is versatile and has an immense range of potential users. 4.
Compact and weather-resistant: naturality and impulsiveness are key in our games, so there's no time for accessories. 5.
Simple Design from a single mould: a simple design lets us keep costs down, and makes the project viable. Following these five points, since its first release in 2007, the sensor has undergone some improvements, leading to different versions ( Figure 2). In the same way, the video games have also evolved, based on the valuable feedback of targeted users (see Section 4). A brief description of the different versions is shown in the following sub-sections.
Sensors 2016, 16, 586 4 of 20 Following these five points, since its first release in 2007, the sensor has undergone some improvements, leading to different versions ( Figure 2). In the same way, the video games have also evolved, based on the valuable feedback of targeted users (see Section 4). A brief description of the different versions is shown in the following sub-sections.

First Version
The first version of the HybridPLAY sensor (released in 2007) basically consisted of an Arduino mini board [28], a three-axis accelerometer and an XBee module for the wireless communication [29]. In this first version, four different models of the sensor were produced according to different elements of the playground: the swing, the slide, the seesaw and the hobbyhorse. In all of them, the housing was composed of a plastic rectangular box (see Figure 2), which was afterwards covered with cardboard ( Figure 3a) and stuck to the elements of the playground with an adhesive Velcro strip (Figure 3b). For the wireless connectivity we use the XBee protocol to create a mesh of sensors that send the information to a Linux laptop provided with an external XBee antenna. The laptop acts as a server with a custom-made Python server software that analyzes the info of the sensors, save the players' scores and forwards the sensor triggers via WiFi to the player.  The first video games were designed by Clara Boj, and programmed by Diego Díaz and Martín Nadal with pygame [30]-a Python game library-and run on a Nokia N850 PDA with the Maemo Linux-based operating system, and produced by Intermediae Matadero Madrid (Madrid, Spain). The image in Figure 4a shows our first video game called Puzzle City, which consists of a set of mini games where the character has to collect the pieces of a puzzle to collect points and fill a gap created in the (virtual) city. When players fill this gap, they discover that the virtual world represents a playground. Each mini game is made up of simple dynamics associated with the movement of the

First Version
The first version of the HybridPLAY sensor (released in 2007) basically consisted of an Arduino mini board [28], a three-axis accelerometer and an XBee module for the wireless communication [29]. In this first version, four different models of the sensor were produced according to different elements of the playground: the swing, the slide, the seesaw and the hobbyhorse. In all of them, the housing was composed of a plastic rectangular box (see Figure 2), which was afterwards covered with cardboard ( Figure 3a) and stuck to the elements of the playground with an adhesive Velcro strip (Figure 3b). For the wireless connectivity we use the XBee protocol to create a mesh of sensors that send the information to a Linux laptop provided with an external XBee antenna. The laptop acts as a server with a custom-made Python server software that analyzes the info of the sensors, save the players' scores and forwards the sensor triggers via WiFi to the player. Following these five points, since its first release in 2007, the sensor has undergone some improvements, leading to different versions ( Figure 2). In the same way, the video games have also evolved, based on the valuable feedback of targeted users (see Section 4). A brief description of the different versions is shown in the following sub-sections.

First Version
The first version of the HybridPLAY sensor (released in 2007) basically consisted of an Arduino mini board [28], a three-axis accelerometer and an XBee module for the wireless communication [29]. In this first version, four different models of the sensor were produced according to different elements of the playground: the swing, the slide, the seesaw and the hobbyhorse. In all of them, the housing was composed of a plastic rectangular box (see Figure 2), which was afterwards covered with cardboard ( Figure 3a) and stuck to the elements of the playground with an adhesive Velcro strip ( Figure 3b). For the wireless connectivity we use the XBee protocol to create a mesh of sensors that send the information to a Linux laptop provided with an external XBee antenna. The laptop acts as a server with a custom-made Python server software that analyzes the info of the sensors, save the players' scores and forwards the sensor triggers via WiFi to the player.  The first video games were designed by Clara Boj, and programmed by Diego Díaz and Martín Nadal with pygame [30]-a Python game library-and run on a Nokia N850 PDA with the Maemo Linux-based operating system, and produced by Intermediae Matadero Madrid (Madrid, Spain). The image in Figure 4a shows our first video game called Puzzle City, which consists of a set of mini games where the character has to collect the pieces of a puzzle to collect points and fill a gap created in the (virtual) city. When players fill this gap, they discover that the virtual world represents a playground. Each mini game is made up of simple dynamics associated with the movement of the The first video games were designed by Clara Boj, and programmed by Diego Díaz and Martín Nadal with pygame [30]-a Python game library-and run on a Nokia N850 PDA with the Maemo Linux-based operating system, and produced by Intermediae Matadero Madrid (Madrid, Spain). The image in Figure 4a shows our first video game called Puzzle City, which consists of a set of mini games where the character has to collect the pieces of a puzzle to collect points and fill a gap created in the (virtual) city. When players fill this gap, they discover that the virtual world represents a playground. Each mini game is made up of simple dynamics associated with the movement of the child in the park, for example ( Figure 4a) the character has to jump between (virtual) clouds by sliding down a (real) slide, where some clouds contain pieces of the puzzle and others none.
The mobile device running the video game was wearable, as it was integrated of a self-designed bracelet (Figure 4b). The system worked with four teams simultaneously, each one with a different bracelet color: blue, orange, red and green.
Sensors 2016, 16, 586 5 of 20 child in the park, for example ( Figure 4a) the character has to jump between (virtual) clouds by sliding down a (real) slide, where some clouds contain pieces of the puzzle and others none. The mobile device running the video game was wearable, as it was integrated of a self-designed bracelet (Figure 4b). The system worked with four teams simultaneously, each one with a different bracelet color: blue, orange, red and green.

Second Version
The second version of the HybridPLAY sensor (released in 2009) was designed by Miguel de las Heras (Hangar, Barcelona, Spain), and was based on the technical specifications of the first version, to which some improvements were added. In particular, all the electronic elements were integrated into a single electronic board and some components were replaced. The main components of the HybridPLAY sensor in this version are: an ATmega168 processor (compatible with Arduino), a mini-USB connector, a three-axis accelerometer, an XBee component, a sound component and RFID reader and antenna. These improvements allowed us to reduce the dimensions of the sensors and the packing box, which also consisted of a rectangular plastic box that was stuck to the elements of the playground with Velcro. The video game (Puzzle City) was improved according of the feedback of kids. For example, we introduced the multiplayer option, where four teams can play the same game together and the laptop, that acts as a server, shows the score of each team, which is identified with a different color (blue, orange, red or green) according to the bracelet worn ( Figure 5). Each team was composed of five members, one of them acting as coordinator, who carried the PDA on a bracelet in such a way that he/she could see the screen at all times to provide precise instructions to the other four team members. These members were running through the park by driving the physical elements in order to control the video games. As the implemented video game is made up of various mini games, the role of the coordinator was swapped among players for each mini game. We also designed new games based on kids' ideas, improved the game speed and cleaned up some computational errors detected in the first version.

Second Version
The second version of the HybridPLAY sensor (released in 2009) was designed by Miguel de las Heras (Hangar, Barcelona, Spain), and was based on the technical specifications of the first version, to which some improvements were added. In particular, all the electronic elements were integrated into a single electronic board and some components were replaced. The main components of the HybridPLAY sensor in this version are: an ATmega168 processor (compatible with Arduino), a mini-USB connector, a three-axis accelerometer, an XBee component, a sound component and RFID reader and antenna. These improvements allowed us to reduce the dimensions of the sensors and the packing box, which also consisted of a rectangular plastic box that was stuck to the elements of the playground with Velcro. The video game (Puzzle City) was improved according of the feedback of kids. For example, we introduced the multiplayer option, where four teams can play the same game together and the laptop, that acts as a server, shows the score of each team, which is identified with a different color (blue, orange, red or green) according to the bracelet worn ( Figure 5). Each team was composed of five members, one of them acting as coordinator, who carried the PDA on a bracelet in such a way that he/she could see the screen at all times to provide precise instructions to the other four team members. These members were running through the park by driving the physical elements in order to control the video games. As the implemented video game is made up of various mini games, the role of the coordinator was swapped among players for each mini game. We also designed new games based on kids' ideas, improved the game speed and cleaned up some computational errors detected in the first version.
such a way that he/she could see the screen at all times to provide precise instructions to the other four team members. These members were running through the park by driving the physical elements in order to control the video games. As the implemented video game is made up of various mini games, the role of the coordinator was swapped among players for each mini game. We also designed new games based on kids' ideas, improved the game speed and cleaned up some computational errors detected in the first version.
(a) (b) (c) Figure 5. Kids looking the score in the server screen (a-c). Details of the bracelet with the integrated RFID tag (b). Figure 5. Kids looking the score in the server screen (a-c). Details of the bracelet with the integrated RFID tag (b).

Third Version
The current version of the HybridPLAY is the third version (released in 2015) [31]. In this case, the electronics of the sensor were also designed by Miguel de las Heras (Hangar, Barcelona, Spain), and consisted on the optimization of the components from the second version. The main components of the HybridPLAY sensor are, in this version ( Figure 6): an ATmega32U4-AU processor (compatible with Arduino), a USB micro connector, a three-axis accelerometer and gyroscope, an infrared proximity sensor, a LED button and a sound component. In this case, the connectivity is achieved with Bluetooth LE (4.1).

Third Version
The current version of the HybridPLAY is the third version (released in 2015) [31]. In this case, the electronics of the sensor were also designed by Miguel de las Heras (Hangar, Barcelona, Spain), and consisted on the optimization of the components from the second version. The main components of the HybridPLAY sensor are, in this version ( Figure 6): an ATmega32U4-AU processor (compatible with Arduino), a USB micro connector, a three-axis accelerometer and gyroscope, an infrared proximity sensor, a LED button and a sound component. In this case, the connectivity is achieved with Bluetooth LE (4.1). The sensor case was designed by Joan Rojeski's design studio with due consideration of the way children interacted with the different elements of the playground. Every decision was supported by prototypes adapted to the strength and dimensions of both children and adults. After several iterations and size adjustments of the clip system, a balanced solution was found, as shown in Figure 7. The sensor case was designed by Joan Rojeski's design studio with due consideration of the way children interacted with the different elements of the playground. Every decision was supported by prototypes adapted to the strength and dimensions of both children and adults. After several iterations and size adjustments of the clip system, a balanced solution was found, as shown in Figure 7. The sensor case was designed by Joan Rojeski's design studio with due consideration of the way children interacted with the different elements of the playground. Every decision was supported by prototypes adapted to the strength and dimensions of both children and adults. After several iterations and size adjustments of the clip system, a balanced solution was found, as shown in Figure 7. Regarding video games, our primary goal for this third version was both to improve the previously developed games (art, design, coding) as well as designing new games for the platform. The games for the third version were developed for both smartphone and tablet devices, and for both Android and iOS platforms, and initially we used the game engines cocos2dx and Unity. In this version the device and the sensors are connected directly via Bluetooth LE, so there is no further need to use the laptop server and the XBee protocol. The app was designed by Clara Boj and Diego Díaz, Regarding video games, our primary goal for this third version was both to improve the previously developed games (art, design, coding) as well as designing new games for the platform. The games for the third version were developed for both smartphone and tablet devices, and for both Android and iOS platforms, and initially we used the game engines cocos2dx and Unity. In this version the device and the sensors are connected directly via Bluetooth LE, so there is no further need to use the laptop server and the XBee protocol. The app was designed by Clara Boj and Diego Díaz, and programmed by Emanuel Mazza and Diego Díaz. This app includes different video games like Space Kids or Puzzle City 2. In Space Kids (Figure 8a), children can clean the galaxy by collecting space debris, explore a space station by balancing on the seesaw, or experience zero gravity by playing on the swings. In Puzzle City 2 ( Figure 8b) children aim to find puzzle pieces to discover what is missing in the city by exploring the neighborhood with his/her team, walking around, and learning what different city spaces are meant for. The app also reinvents classic video games such as Pac-Man and Pong. Kids are aimed to play Pong on the swing, or help Pac-Man to escape moving their body on the seesaw. The physical movements of the elements of the park, triggered by children, and their representation in video games, have a direct correlation to facilitate and optimize the dynamics of the game, with the aim that these dynamics based on movements become more intuitive, respecting and helping children to develop their spatial logic.  The physical movements of the elements of the park, triggered by children, and their representation in video games, have a direct correlation to facilitate and optimize the dynamics of the game, with the aim that these dynamics based on movements become more intuitive, respecting and helping children to develop their spatial logic.
For instance, when a child tilts the see-saw to the right, the character in the game moves to the right, and when he tilts it forward, the character moves up, etc. In the case of the slide, the character jumps when the child slides. In the case of swing, the character moves synchronized with its movements, drawing a circumference arc. Therefore, on the one hand, the actions that the child has to perform in the playground are the same or very similar to those that he would do to play without electrical devices and, on the other hand, the physical actions have a logical and intuitive correlation in the video game. In Table 1, a summary of the technical specifications for the three versions of the sensor is listed. As it can be seen, each version improves the capabilities of the former.

Interface Design
In the following sub-sections, the technical specifications of the interface design regarding Bluetooth communication, frame rate, sensoring and user interaction are described in detail for the current version of the system-the third version. Within this, we aim at providing detailed information of the input/output parameters of the sensors and how the interaction is produced according to the different elements of the park.

Bluetooth Communication
In order to communicate the sensor with the mobile devices, the Bluetooth Low Energy (BLE) or Bluetooth Smart protocols are used, both belonging to Bluetooth 4.0. The chip used is the RN4020, which is internally programmed to accommodate a private GATT service with three characteristics: NOTIFY (11 bytes), WRITE (2 bytes) and WRITE (2 bytes). In the characteristic NOTIFY, a string is recorded in hexadecimal form, which contains the following values from the sensors: IR (1 byte), BATTERY (1 byte), TILT (1 byte) and QUATERNIONS (8 bytes), where IR stands for the infrared proximity sensor. All of these values are normalized in integer values. In the case of IR and BATTERY, the values go from 0 to 100; TILT is a Boolean, so it can be 0 or 1; and QUATERNIONS go from 0 to 255.
The primary difference with a standard communication through the serial port (used for example in the classic Bluetooth 2.0) is that, in the case of using private services and features, there is already a Sensors 2016, 16, 586 9 of 21 "container" prepared for the variable, which is directly accessible. By contrast, with a standard reading of the serial port, the data must be packaged and temporized to avoid overflow. For instance, it is common in a standard script to add a header to the message and to add a control on the reception in order to start reading only after receiving the header; this can lead to eventual jumps in reading because, if mistakenly the header is not automatically read, the rest of the message is lost until the next refresh. With the private services of the BLE, this problem disappears.

Frame Rate
Regarding the electronics, the firmware internal clock is set to 25 Hz, which corresponds to 25 refreshes per second or a single refresh every 40 ms. The refresh rate is the same for all sensors. Regarding the application, a new reading of the values recorded in the feature NOTIFY (which records all sensor data) is performed at each update cycle.

Sensoring
To retrieve the rotations of the sensor, the quaternion system was used, which is represented by the parameters q 0 , q 1 , q 2 , q 3 . Though less intuitive than Euler angles, quaternions provide an alternative measurement technique that does not suffer from gimbal lock, which arise when two of the three axes are aligned and a degree of freedom is lost. The relationship between quaternions and the Euler angles φ (roll), θ (pitch) and ψ (yaw) can be defined as follows: rcsin p2 pq 0 q 2´q3 q 1 qq where the resulting angular values are between´180˝and +180 for φ and ψ, and between´90 and +90 for θ. The TILT parameter is internally detected by the firmware after a calibration (some consecutive lectures of the accelerometer) when the sensor is initialized. This calibration value is continuously compared with the data of the gyroscope and the accelerometer to detect if they go outside a defined range, in which case a sudden movement is detected. The accuracy of the TILT depends both on the sensor and on the defined tolerance level. The infrared proximity sensor that is currently being used, the GP2Y0E02B, has a detection range between 3 and 100 cm. Beyond this range, the sensor is set to inactive.

Springy Elements
When attaching the sensor on any element of the playground that is provided with one or more springs (e.g., seesaw or hobbyhorse), the behavior of the sensor is similar to the traditional joysticks. Initially, a simple but effective calibration process is carried out (Figure 9). First, the sensor is coupled in any of the orientations indicated in the app, and the user selects in the app the chosen position.

Springy Elements
When attaching the sensor on any element of the playground that is provided with one or more springs (e.g., seesaw or hobbyhorse), the behavior of the sensor is similar to the traditional joysticks. Initially, a simple but effective calibration process is carried out (Figure 9). First, the sensor is coupled in any of the orientations indicated in the app, and the user selects in the app the chosen position. Once this configuration is finished, when starting the game, the system automatically calibrates the sensor with respect to the inclination it has. For a proper calibration, it is necessary that the element of the park is as still as possible.
Thanks to its sensor, two general types of motion can be detected: tilt and inclination. With the tilt, sudden movements are detected, and the seesaw/hobbyhorse game dynamics associated with it are established, such as increasing energy or the value of a variable of the player. For example, in the case of Moskis, the fly begins its flight when tilting the element of the park (Figure 10). Once this configuration is finished, when starting the game, the system automatically calibrates the sensor with respect to the inclination it has. For a proper calibration, it is necessary that the element of the park is as still as possible.
Thanks to its sensor, two general types of motion can be detected: tilt and inclination. With the tilt, sudden movements are detected, and the seesaw/hobbyhorse game dynamics associated with it are established, such as increasing energy or the value of a variable of the player. For example, in the case of Moskis, the fly begins its flight when tilting the element of the park (Figure 10). The 4-way movement is more complex, and is detected when there is a significant variation in ϕ (tilt left-right) and θ (tilt forward-back) with respect to the values recorded in the initial calibration. These movements are used in video games to control the direction of the user, often on a twodimensional plane in an aerial view, such as in the mazes of the adaption of the classic arcade video game Pac Man (Figure 11). The 4-way movement is more complex, and is detected when there is a significant variation in φ (tilt left-right) and θ (tilt forward-back) with respect to the values recorded in the initial calibration. These movements are used in video games to control the direction of the user, often on a two-dimensional plane in an aerial view, such as in the mazes of the adaption of the classic arcade video game Pac Man (Figure 11). The 4-way movement is more complex, and is detected when there is a significant variation in ϕ (tilt left-right) and θ (tilt forward-back) with respect to the values recorded in the initial calibration. These movements are used in video games to control the direction of the user, often on a twodimensional plane in an aerial view, such as in the mazes of the adaption of the classic arcade video game Pac Man (Figure 11). The effect of the inclination in the game can be digital (activating a predefined movement speed) or analog, so that the speed of the associated action is related to the value of the inclination angle.

The Swing
In the case of the swing, the sensor must be coupled to the base thereof facing the floor, so that it can be detected. When the child sways, the sensor recognizes everytime that the swing is closer to the ground (Figure 12). In this way, the swinging speed can be calculated, since the cutting period is higher or lower depending on it. The effect of the inclination in the game can be digital (activating a predefined movement speed) or analog, so that the speed of the associated action is related to the value of the inclination angle.

The Swing
In the case of the swing, the sensor must be coupled to the base thereof facing the floor, so that it can be detected. When the child sways, the sensor recognizes everytime that the swing is closer to the ground (Figure 12). In this way, the swinging speed can be calculated, since the cutting period is higher or lower depending on it. These movements can be reflected in the actions of the app in different ways, for example to control the speed of the protagonist or to also control its position, moving from left to right of the screen, in a linear motion, or forming an arc. An example is depicted in Figure 13, where the movements of the swing produce an interaction in the video game HybridEDU (in this example, intended to play a piano). These movements can be reflected in the actions of the app in different ways, for example to control the speed of the protagonist or to also control its position, moving from left to right of the screen, in a linear motion, or forming an arc. An example is depicted in Figure 13, where the movements of the swing produce an interaction in the video game HybridEDU (in this example, intended to play a piano).
These movements can be reflected in the actions of the app in different ways, for example to control the speed of the protagonist or to also control its position, moving from left to right of the screen, in a linear motion, or forming an arc. An example is depicted in Figure 13, where the movements of the swing produce an interaction in the video game HybridEDU (in this example, intended to play a piano).

The Slide
In this case the sensor must be attached to the side of the slide, so that the infrared proximity sensor is facing the child, sensing his presence when sitting at the top of the slide. To ensure that the child is waiting to pounce, the app detects whether the infrared proximity sensor has been cut for at least 2 s. Once the presence of the child has been detected, a signal appears in the game indicating that the system is ready to detect his jumping ( Figure 14).

The Slide
In this case the sensor must be attached to the side of the slide, so that the infrared proximity sensor is facing the child, sensing his presence when sitting at the top of the slide. To ensure that the child is waiting to pounce, the app detects whether the infrared proximity sensor has been cut for at least 2 s. Once the presence of the child has been detected, a signal appears in the game indicating that the system is ready to detect his jumping ( Figure 14). These movements can be reflected in the actions of the app in different ways, for example to control the speed of the protagonist or to also control its position, moving from left to right of the screen, in a linear motion, or forming an arc. An example is depicted in Figure 13, where the movements of the swing produce an interaction in the video game HybridEDU (in this example, intended to play a piano).

The Slide
In this case the sensor must be attached to the side of the slide, so that the infrared proximity sensor is facing the child, sensing his presence when sitting at the top of the slide. To ensure that the child is waiting to pounce, the app detects whether the infrared proximity sensor has been cut for at least 2 s. Once the presence of the child has been detected, a signal appears in the game indicating that the system is ready to detect his jumping ( Figure 14). Figure 14. Schematic of the interaction produced by placing the sensor on the slide. The system detects that the child is waiting for jumping. Figure 14. Schematic of the interaction produced by placing the sensor on the slide. The system detects that the child is waiting for jumping.
Regarding to the child jumping to the slide, it is only detected the instant when it occurs, but not the speed or time sliding down the slide. Other actions may be associated with the activation of any event, such as a shot or the game's protagonist jump.

Other Possibilities of User Interaction
The games designed for HybridPLAY can also use the classical mechanisms of interaction provided on smartphones, such as the interaction with the screen through the touch (single touch or multi touch) or the internal accelerometers. The actions associated with these interactions are the same as in other video games for mobile devices, but are of particular interest and relevance as they complement those of the HybridPLAY device, as the child with the role of coordinator has to order the actions to be taken, at the same time that he/she interacts with the game.

Evaluation and Discussion
Since its beginnings in 2007, the HybridPLAY system has been tested and exhibit in many different places, from artistic exhibitions to conferences and workshops. Altogether, the HybridPLAY system has been tested by around 1500 children and 400 parents (estimated values). Based on the feedback of these trials, the sensor has evolved from its first to the current version. The complete list where HybridPLAY has been tested/exhibited up to date is provided in the following points, according to its different versions. The approximate number of people (children and adults) that tested the system is also given. The first version was exhibited and/or tested in:

Workshops
The performed workshops with group of children have had an approximate duration of two hours each, and have been organized as indicated in Table 2.
The experience of the workshops was very positive. From the first moment, children were quite engaged with the proposed games, and enjoyed the system very much. However, despite the great success of the system in the first version, we found some aspects that needed to be improved, which were based on our personal observations and the feedback from parents and children.
Regarding to the participation of children, we noticed that children between 6 and 10 years old responded better to the system, while children under 6 didn't have enough strength to move the spring elements and also experienced some difficulty in understanding the dynamics of the games children over 10 tended to feel very attracted to the system, but the available games seemed too simple to them. Besides, some children under 8 had difficulty discerning between left and right, and also in being to swing on the swings without the help of another child or adult. Regarding the task of coordinator, children performed differently according to their personal profiles: while some of them felt more comfortable doing this task, others (more introverted) had a harder time at first, but soon developed well in this task. We also found no integration problems with the other children of the park that were not playing with HybridPLAY, so we can say that the system does not interfere with the normal use of the park by other children. Additionally, after finishing the workshops, many children continued playing in the park imagining they were doing that with HybridPLAY, and also brought their own ideas for games to take place, and thus we can conclude that the system encourages imagination and creativity. Also, parents were easily engaged into the dynamics of the game, and so HybridPLAY functions as an integrator of family activity at playgrounds.

min
Brief talk relating to the games in the playground. Some questions arise, such as: ‚ Do you usually play in the playground? ‚ What are your favourite elements of the playground? ‚ What element do you like the least? ‚ Is there any missing element which you would like to play with? Which one? ‚ Do you find funny the see-saw, the swing, etc.?

min
Brief talk relating to the workshop, explaining: ‚ The games proposed in HybridPLAY ‚ The role of children as avatars ‚ The responsibility of children regarding each other and the devices that they carry during the game ‚ The organization into two teams 10 min The elements of the game: the way the HybridPLAY sensors are integrated in the different elements of the playground and how do they work.

min
Creation of gaming teams: ‚ Four groups of 3 to 4 players ‚ Two groups of observers These groups rotate, in such a way that every children becomes a player and an observer. Each group of players has a game coordinador, which is the child that holds the smartphone. For every mini game, the game responsibility changes, so all the kids will hold the smartphone and coordinate the team. The sensors are handed to the children.

min
Game experience: Children freely play with the HybridPLAY platform integrated in the elements of the playground. Observation. The observers do not participate in the game, but only see the actions of their colleagues.

min
Group rotation. Players become observers and vice versa. Besides these good results, there were other issues and technical aspects that could be improved. For instance, regarding the game dynamics, it would be interesting to compete between teams for a global gaming session in order to increase user engagement and game capabilities. Regarding the technology itself, we found that the technological system used in this first version was not very portable and lacked complete autonomy (e.g., short battery life). Regarding the proposed video games, it would be interesting to add other elements, such as animations. We were also conscious that the videogames responses were slightly slow (mainly due to the mobile technology) and that some bugs had to be corrected.
All the aforementioned negative aspects were tackled in the second version of HybridPLAY. The autonomy and portability of the system was greatly improved. In this regards, we constructed a portable cart to transport all the equipment, which was equipped with a car battery and an inverter to provide power to the laptop and recharge the batteries of the sensors and PDAs. When the car was unfolded, a 24-inch flat screen showed the players the status of their game, the mini game they were playing and the points scored by each of the four teams. This display was very well received as children continually ran to the car to see the status of the game.
The workshops carried out with the second version of the system also had great acceptance by children and parents. Additionally, we found that the game dynamics were more fluent due to the improvements that were made. For instance, the improvement of the battery life and the absence of inconsistence errors in the video games was deterministic. In this way, the game was not interrupted at any time, leading to a greater user engagement. Also, the addition of animations in the video games and the possibility to compete among teams was greatly accepted by both children and parents, as the gaming possibilities augmented, as well as the teamwork. However, other issues were found which motivated other modifications and improvements, leading to the third version of the system.
In the third version, we included more complete and sophisticated dynamics in the videogames, as well as sound output and more attractive animations, also increasing the speed response. Regarding to the technology, the autonomy and portability of the system was further increased (allowing 20 h of continuous play), while the server computer was eliminated, and hence also the need to use the car with the car battery and the inverter.
The improvements made in the third version lead to a pre-industrial system, with all the aforementioned encountered issues addressed. The feedback after the last workshop carried out to date (TEDx Barcelona) make us think that HybridPLAY has great potential in the entertainment, education and fitness markets, and that the targeted users are not restricted to children, but also e.g., parents are engaged in the game.

Exhibitions
The HybridPLAY system was presented at some artistic exhibitions, attracting the curiosity of the general public. In Figure 15b the system is depicted at the Techformance Festival (Murcia, Spain, 2009). As this was version 2, the cart was used, which was designed with the purpose of augmenting the autonomy of the overall system. The cart was provided with a car battery and an inverter, that allowed us to increase the autonomy of the system from 2 to 6 h and also to recharge the PDAs. This was relevant for the exhibitions, where the system was on display for the whole day.
The cart had a set of drawers covered with foam to absorb impacts and to transport tools, the Velcro, sensors, batteries, inverter, etc. It was equipped with two doors that could be extended, forming a table and, when closed, it was compact enough and highly resistant to transport all the material, so it could be easily transported to different exhibition venues. forming a table and, when closed, it was compact enough and highly resistant to transport all the material, so it could be easily transported to different exhibition venues.

Quantitative Evaluation
A quantitative user evaluation of the HybridPLAY system was carried out during the Workshop La Fira d'Oci Infantil i Juvenil de Nadal, (Castellón, Spain, 2016) with the third version of the system. Around 150 people tested the system at the workshop, from which a total of 87 acceded to fill out a pair of questionnaires, that were related to the usability of the system and the individual's satisfaction with the proposed activity, where the System Usability Scale (SUS) [32] was chosen to measure usability.
The participants were both adults (32 participants) and children (55 participants). The adults that participated in the survey were the parents of children, who often took the role of coordinators. In Table 3, the socio-demographic data of these two groups is provided, showing the mean, standard deviation (s.d.) and minimum and maximum age of the participants. Other parameters used are gender and how often they use video games, where the range 0-4 for the frequency (f.) means: 0: never or almost never; 4: every day. Most of the children participating in the questionnaires fulfilled them with the help of their parents and/or other adults supervising the activity. As we have detected slightly different results in the questionnaires regarding adults and children, so we present the results divided into these two groups. We have, however, not detected relevant differences regarding gender and video game use frequency. The results of the SUS questionnaire are listed in Table 4, where the obtained scores are provided for the two considered groups, adults and children. In questions 1 to 10 the range 0-4 means: 0: strongly disagree, 4: strongly agree. The values of the SUS score, however, range from 0 to 100, meaning 100 the best imaginable result. For the adults group, this score reaches 84.92 points; for the children group, the score is slightly better, reaching 91.23 points. These values can be considered good (for adults) and excellent (for children) on the scale of scores provided by the questionnaire and taking into account the fact a minimum score of 68 would be deemed acceptable for a tool [33,34].

Quantitative Evaluation
A quantitative user evaluation of the HybridPLAY system was carried out during the Workshop La Fira d'Oci Infantil i Juvenil de Nadal, (Castellón, Spain, 2016) with the third version of the system. Around 150 people tested the system at the workshop, from which a total of 87 acceded to fill out a pair of questionnaires, that were related to the usability of the system and the individual's satisfaction with the proposed activity, where the System Usability Scale (SUS) [32] was chosen to measure usability.
The participants were both adults (32 participants) and children (55 participants). The adults that participated in the survey were the parents of children, who often took the role of coordinators. In Table 3, the socio-demographic data of these two groups is provided, showing the mean, standard deviation (s.d.) and minimum and maximum age of the participants. Other parameters used are gender and how often they use video games, where the range 0-4 for the frequency (f.) means: 0: never or almost never; 4: every day. Most of the children participating in the questionnaires fulfilled them with the help of their parents and/or other adults supervising the activity. As we have detected slightly different results in the questionnaires regarding adults and children, so we present the results divided into these two groups. We have, however, not detected relevant differences regarding gender and video game use frequency. The results of the SUS questionnaire are listed in Table 4, where the obtained scores are provided for the two considered groups, adults and children. In questions 1 to 10 the range 0-4 means: 0: strongly disagree, 4: strongly agree. The values of the SUS score, however, range from 0 to 100, meaning 100 the best imaginable result. For the adults group, this score reaches 84.92 points; for the children group, the score is slightly better, reaching 91.23 points. These values can be considered good (for adults) and excellent (for children) on the scale of scores provided by the questionnaire and taking into account the fact a minimum score of 68 would be deemed acceptable for a tool [33,34]. The results of the individuals' satisfaction questionnaires are given in Table 5 for the two considered groups. The scores also range from 0 to 4, meaning: 0: strongly disagree, 4: strongly agree. As it can be seen, results are very satisfactory. As a general observation, it can be seen that the adults is more critical, as slightly worse scores are obtained for almost all the questions of the two questionnaires. Besides, the obtained results are excellent or almost excellent, what we believe is the consequence of the different improvements carried out for the HybridPLAY system, after listening to users' feedback during the many organized workshops and exhibitions during the last years.

Extended Uses of HybridPLAY
Although our main targeted audience are children, the HybridPLAY sensor can be also used by other users such as seniors. In Figure 16, an example is shown with the sensors placed at other elements of the playground that are intended to do exercise. In this case, the video games were adapted, in such a way that we simplified the graphics and game dynamics by using basic shapes and colors. We would like to stress the potential value of HybridPLAY to help the physical and verbal synchronization of seniors by motivating their mental activation and improving their response time to external stimuli. The versatile character of HybridPLAY means that it is not restricted to be outdoor use, and indoor activities can also be performed. In this regards, HybridPLAY has also been tested at the classroom level ( Figure 17) and extra school activities, where kids can enjoy and learn at the same time they are playing with HybridPLAY. We have developed a set of basic games with educational contents such as maths, where kids can learn basic addition and multiplication operations. So far, we have received a very positive feedback from this experience, maybe due to the fact that the physical activity helps the mental activation and the subsequent learning process. The video games can be also customized, in such a way that the educational sector can benefit from HybridPLAY, including the education of kids with special needs. In this regards we have developed some basic experiences with kids with diminished mental capabilities and we have realized that in some cases, after properly adjusting the level of the game, our system can be very positive for their mental development. The versatile character of HybridPLAY means that it is not restricted to be outdoor use, and indoor activities can also be performed. In this regards, HybridPLAY has also been tested at the classroom level ( Figure 17) and extra school activities, where kids can enjoy and learn at the same time they are playing with HybridPLAY. We have developed a set of basic games with educational contents such as maths, where kids can learn basic addition and multiplication operations. So far, we have received a very positive feedback from this experience, maybe due to the fact that the physical activity helps the mental activation and the subsequent learning process. The versatile character of HybridPLAY means that it is not restricted to be outdoor use, and indoor activities can also be performed. In this regards, HybridPLAY has also been tested at the classroom level (Figure 17) and extra school activities, where kids can enjoy and learn at the same time they are playing with HybridPLAY. We have developed a set of basic games with educational contents such as maths, where kids can learn basic addition and multiplication operations. So far, we have received a very positive feedback from this experience, maybe due to the fact that the physical activity helps the mental activation and the subsequent learning process. The video games can be also customized, in such a way that the educational sector can benefit from HybridPLAY, including the education of kids with special needs. In this regards we have developed some basic experiences with kids with diminished mental capabilities and we have realized that in some cases, after properly adjusting the level of the game, our system can be very positive for their mental development. The video games can be also customized, in such a way that the educational sector can benefit from HybridPLAY, including the education of kids with special needs. In this regards we have developed some basic experiences with kids with diminished mental capabilities and we have realized that in some cases, after properly adjusting the level of the game, our system can be very positive for their mental development.

Conclusions and Further Work
In this paper we have presented HybridPLAY, a new platform composed of sensor units and video games that fosters outdoor physical activity, verbal communication and teamwork. We have described the work performed to date, both at the technological and at the evaluation (user-related issues) level, that has led us to achieve a robust version of both the sensor and the video games. The results of a quantitative evaluation of the system carried out for its third version are excellent, what we believe is a consequence of the different improvements carried out for the HybridPLAY system, after responding to users' feedback during the many organized workshops and exhibitions attended during the last years.
We also have demonstrated the versatile character of HybridPLAY. Though initially intended to foster outdoor physical activity in playgrounds, where the main targeted users are children, HybridPLAY can also be used by other audiences (e.g., seniors) and can be used indoors. Additionally, the video games can be customized in order to provide benefits in the educational sector, including the education of people with special needs.
HybridPLAY is a platform that started as an artistic initiative to address actual problems, that however has today reached great acceptance by the public and has become the primary product of Lalalab Projects. With a view the mass production and commercialization of the HybridPLAY platform, some steps are still required. In particular, regarding the hardware, we intend to reduce the dimension of the sensor; regarding to the software, we intend to build a platform that allow users to design their own games. We also intend to introduce the sensor in the entertainment, education and fitness markets, possibly customizing some of the features of HybridPLAY (e.g., video games) for the different sectors. Additionally, we believe that HybridPLAY can be used as a complementary tool to fight overweight and obesity in children, as it fosters physical activity. In order to demonstrate this claim, we plan to provide a group of children with our technology for two or three months, to track their physical activity and to evaluate the use they make of it.