Pages: pp. 45-55
Abstract—Many studies have attested to the efficacy of integrating innovative methods of teaching computing and engineering, especially for introductory students and at the K-12 level. As an example, robots have been successfully used to teach a wide range of subjects, from introductory programming to artificial intelligence. As a discipline, wearable computing is seen to be practical, yet futuristic and exciting, and it captures the attention and interest of people who might not otherwise be drawn to technology. Recent developments in this field have also raised the possibility of moving wearable computing construction within the reach of hobbyists and novices. However, there still exist substantial obstacles toward the adoption of wearable computing into education. This paper presents a framework with the objective of facilitating the integration of wearable computing into outreach and introductory computer science and engineering education. We also present a comprehensive evaluation of our platform, including a comparison with the current practice of sewing-based wearable computing.
Index Terms—Computer and information science education, programming environments/construction tools, wearable computers and body area networks.
The number of students who are interested in technology, and subsequently, the enrollment at engineering and computer science programs across the world has been declining for the last few years. To stimulate and hold the interest of students, much effort has gone into research and development of innovative methods of teaching. For example, many interactive multimedia-based programming environments, such as Scratch [ 1], Alice [ 2], and Logo, have been developed, mostly for use by K-12 children. There has also been a lot of development in educational robotics for wide variety of subjects and age ranges, from introductory programming to artificial intelligence to physics, at both the K-12 and university level.
Compared with purely virtual environments, robotics has the advantage of being tangible—in that students are required to build their own construction and program it to perform certain tasks. This is beneficial to the learning process as there is some evidence that perception and cognition are closely interlinked [ 3], [ 4], [ 5], [ 6], [ 7], and the physical manipulation of tangible objects might support more effective or more natural learning [ 8], [ 9], [ 10]. In addition, collaborative learning and construction are usually more feasible with physical objects than with virtual interfaces [ 11]. As a result, robotics is very widely used at all levels of education and there is a lot of work attesting to its effectiveness. However, because of the nature of the domain, the construction of robots requires a significant amount of spatial and mechanical awareness. For example, even a task as simple as getting a robot to make a tight turn requires some consideration of friction and gears. Students who do not possess good spatial or mechanical "sense" usually get very frustrated with the construction process, even before they have gotten to the stage where they can get the robot to do some interesting tasks!
We believe that the emerging field of wearable computing and e-textiles holds much promise in the educational computing arena. After many years of research, wearable computing is showing signs of leaving the research laboratory and emerging into the world. Examples of typical consumer applications include body function monitoring products such as the Nike Bluetooth shoe that monitors the user's pace while running, as well as jackets with integrated iPod controls. These products have raised the awareness of the general public toward this field, with the result that wearable computing is perceived as being futuristic and exciting. In addition, the deployment of wearable electronics in public events such as the opening ceremony in the Beijing Olympics has injected a strong esthetic and creative component. These two characteristics help to appeal to a different demographic of young people, who might not be otherwise drawn to computers or engineering. As an educational tool, a course or a kit in wearable computing would serve to help children exercise their creativity and innovation, at the same time exposing them to elements of technology, computing, and logic through the deployment and integration of sensors and actuators.
Until recently, the realm of wearable computing was accessible only to experts with the resources and the knowledge to build customized wearable computing constructions. The development of the Lilypad Arduino [ 12] was designed to "lower the bar" for wearable computing by bringing e-textiles within the reach of the hobbyist and the classroom. The Lilypad series includes a microcontroller board, sensors, and actuators that can be attached to clothing using readily available tools such as sewing needles and easily obtained materials such as conductive thread. In addition, techniques were developed to construct electronic parts such as PCBs and sockets from fabric-based materials [ 13].
The usability of these materials was demonstrated in six science and e-textile outreach workshops [ 14], as well as in the Eduwear project in Europe [ 15]. However, there still exist a number of problems with deploying wearable computing in education. To begin with, the nature of educational computing requires robust tools with a large error tolerance and a low entry barrier, as they will be used by novices with a large range of skill sets and also be subjected to quite a lot of wear-and-tear. In addition, as the trial-and-error process is paramount in the learning process, it is imperative that educational computing tools be easily debuggable and reconfigurable, so that students will not be afraid to try and make mistakes.
Currently available state-of-the-art technology still faces a number of challenges where these requirements are concerned. For instance, we can consider the task of attaching electronic components such as sensors and LED lights to a garment. Current practice requires that these components be hand-sewn onto the fabric using conductive thread, which is also used to create the signal traces on the garment using lines of hand or machine-sewn stitches. This process is labor-intensive and requires considerable skill to ensure a secure connection between the component and the thread. Undoing a mistake requires considerable labor as stitches have to be picked out one by one. Previous work [ 16] reports that oftentimes, students spend too much time on sewing and not enough time on programming, which is suboptimal when the objective is to expose and teach them about computing and engineering.
In addition to the above, significant problems also exist with the constraints and imperfections of the construction materials. The conductive thread that is used for securing components to the fabric as well as providing a conductive pathway for signals and power has a nonnegligible resistance, which can cause the voltage of the power supply to drop appreciably over distances of as little as a few inches. This has been documented to cause some components to malfunction [ 17], and as a result, imposes further variable factors into the situation. Aside from that, conductive thread also frays easily, causing adjacent traces to contact and short-circuit each other.
Finally, the choice of programming language also poses difficulties to the widespread adoption of wearable computing in education. The state-of-the-art Arduino Lilypad is programmed in a dialect of C, which is not a language that is very easily picked up by beginners, especially by children. Previous work [ 14] has documented the frustrations that children felt with programming their wearable computing construction and also expressed the need for user-friendly programming languages and environments for working with e-textiles and wearable computing.
As a result of these limitations, the educational usage of wearable computing has so far been limited to small-scale workshops of at most around 10 children, run by experts in the field. These obstacles will have to be overcome before wearable computing can be deployed in K-12 outreach workshops or in introductory computer science or engineering courses, akin to the widespread deployment of robotics in educational computing.
In order to overcome the existing obstacles to integrating wearable computing into educational computing, we propose a solution that will address the above problems. Our solution consists of a hardware/physical component: a construction platform for wearable computing creations, as well as a software component: a hybrid text-graphical programming environment. In addition, we will also propose a sample syllabus for a 5-day outreach workshop that effectively guides students through the basic concepts involved in wearable computing.
Our platform will be evaluated through practice. In the evaluation section, we will present results and findings from a sample summer camp in which this platform was deployed.
One of the biggest attractions of using robotics or wearable computing in educational computing is the tangible factor. Previous work [ 7] argues that the use of physical materials in a learning task enhances absorption of concepts and knowledge more readily than if purely virtual objects are used. From experience, students derive a greater sense of reward and satisfaction when successfully constructing a physical object as compared to constructing a virtual one.
Given this, a wearable computing construction platform deserves special attention. In order to adequately support the demands of educational computing, the platform would have to satisfy the following requirements:
The TeeBoard [ 18], a robust and reconfigurable t-shirt integrated with a breadboard, was designed specifically to fulfill the above requirements. Inspired by the solderless breadboards that are commonly used in prototyping and teaching of electronics, it makes use of materials used in wearable garments, and avoids electrical components such as wires, plastic insulation, and solder as much as possible. At the same time, it aims to achieve performance as close as possible to those of its electrical equivalents, enabling designers to focus on designing intelligent behavior into the garment, rather than having to worry excessively about material imperfections.
The platform is also usable by people with minimal sewing and electronic skills, making it suitable for use by students with diverse skills and backgrounds. The system is easily reconfigurable and debuggable, encouraging rapid prototyping and experimentation, and eventual customization. Finally, it is sufficiently versatile to support different kinds of microprocessors, sensors, and actuators.
The TeeBoard makes use of a conductive fabric, Shield-it [ 19], to construct conductive paths on a garment, such as a TeeShirt. It offers much lower resistance than conductive threads and negligible fraying. It has a nonconductive hot-melt adhesive backing on one side. Hence, conductive paths can be ironed over another without short-circuiting. Fig. 1 illustrates a radial pattern created using the conductive fabric. Each line represents a conductive path. The red line and its adjacent black line represent power and ground paths. Each dot is a connective snap button. A microcontroller can then be placed at the "socket location" at the center, to control sensors and actuators placed at strategic locations on the TeeShirt. A set of paths extend to the back of the shirt and the sleeves for flexibility.
Figure Fig. 1. TeeBoard conductive strip pattern. Each line represents a conductive strip. The red line and the adjacent black one are reserved as bus strips that provide the electronic components with the power and ground supply. Each dot represents a connective snap button.
Conductive fabric is also used to create ribbon wires, the fabric equivalent of jumper wires. A thin strip of conductive fabric is threaded through two lengths of ribbon sewn together along their long edges, as shown in Fig. 2.
Figure Fig. 2. Ribbon wires being deployed on the Tee-Board.
In place of sewing and soldering, we repurposed snap buttons for connection between conductive paths, ribbon wires, and electronic components (see Fig. 3). The gripper-style snap buttons can be easily fixed with a hand-operated button gun. Good sewing skills are not required. When a button is secured to the fabric, a secure and stable connection can be easily and reliably made, a big improvement over sew-on snap buttons. These buttons are robust and stand up well to repeated connect/disconnect cycles common in prototyping and customization.
Figure Fig. 3. The snap button connective interface.
We have also devised a doublesnap button in which a snap button can be snapped onto another snap button, enabling distribution of a signal onto multiple paths. Using the doublesnap button, the TeeBoard can also accommodate connections on both sides of a garment, This enables actuators which were meant to be seen, such as LEDs, to be placed on the outside of a garment while other components such as wires and microcontrollers to be placed on the inside, such as in Fig. 4.
Figure Fig. 4. (a) The front side of the finished TeeBoard, with a demonstration circuit illustrating the ability to accommodate circuits on both sides of the garment. (b) The inside of the TeeBoard has the electronic components attached, including the microcontroller board in the center "socket" and a number of connecting wires. The outside of the TeeBoard has a light sensor attached on the top right side (circled) and six LEDs (four were wrapped in colored gauze to simulate flower decorations and two were multicolor LEDs).
The doublesnap buttons are also used to provide easy access to a power supply. A small pocket was constructed in an unobstructive location on the inside of the TeeBoard, which was just big enough to hold a small rechargeable lithium battery, or an AA-size alkaline battery. Wires, snapbuttons, and a switch were soldered onto the terminals of a rechargeable battery, which was then snapped onto the appropriate bus strips.
The TeeBoard fulfills part of our requirements by providing a minimal-skill, flexible platform that allows students with diverse skills to build their own wearable computing outfit by attaching electronic components to the garment. For the garment to be interactive, it needs to be programmed to react to signals received by its sensors and activate its actuators.
Current state-of-the-art programming for wearable computing usually relies on the Arduino development environment, which requires users to program in a subset of the C programming language. This creates problems when the participants are novices or beginners in computer science, as it requires them to learn the logic of programming together with the syntactic constraints of a programming language.
To fulfill the software and programming environment requirements, we developed BrickLayer [ 20], a hybrid graphical-textual programming development system designed for novice programmers, who have little or no programming experience. Fig. 5 shows this programming environment, where users drag blocks (or bricks) which represent programming constructs and drop them onto an "icon-laying" area to create a "wall." To give students an idea of the language that a computer "talks" in, the corresponding source code is instantly generated and displayed in another frame for their viewing.
Figure Fig. 5. BrickLayer's interface consists of a block area (left), construction area (middle), and source code (right). The block area is composed of constructs. The construction area is provided for dropping blocks. The source code is a place for the specific code generation.
As BrickLayer is designed to be a teaching tool, we needed it to accommodate students who have no programming background. To aid the learning process, the BrickLayer environment was developed with three types of constructs: intelligent, fundamental, and user-defined. Intelligent constructs provide checks and balances which protect the user from making careless mistakes. One of the most basic tasks involves controlling the output pins on the microcontroller to perform simple tasks, such as turning LEDs on and off. The way that the microcontroller is set up, a particular pin needs to be "declared" as input or output before it can be used. Since we noticed in our early tests that users often forgot to perform this particular step, BrickLayer automatically inserts statements for setting the behavior and the voltage of a pin into the particular areas of the source code once the required boxes have been dragged and dropped into the construction area.
Apart from the intelligent constructs, BrickLayer also includes fundamental constructs such as conditions, loops, and variables. In many graphical programming environments such as Scratch and NXT-G [ 22], students are overprotected by the constraints of the programming environment. For example, the if statement icon in Scratch encompasses the entire conditional block, which guides and protects the user in placing statements that are under the scope of the conditional statement. This protects the user from his or her own mistakes, but it is possible that students will become overreliant upon the interface and not remember the structures of the programmatic construct when they move to more advanced programming environments. In contrast, BrickLayer requires the student to drag the drop the separate "If," "Then," and "End If" components of a conditional statement individually into the construction area ( Fig. 6). This gives them a more realistic view of programming concepts and constructs, and may make it easier for them to make the switch to conventional textual programming later on.
Figure Fig. 6. Fundamental constructs in Scratch (left), BrickLayer (middle), and C programming (right).
One problem with graphics-based programming environments is that it is often difficult for users to determine the scope of programming constructs. For example, consider the case of a nested if statement that consists of an if statement inside the scope of another if-else statement. If only graphical icons are used, it is often not easy for users to trace the scope of the conditional statement from the dropped blocks. In order to aid the user, BrickLayer assigns a unique color to each conditional or looping block, and all icons within that block are rendered with that color. That allows the user to more easily trace and debug their program (see Fig. 5).
The final type of programming constructs is the user-defined function that we do not expect beginner students to be able to construct. For instance, in order to recognize time-series signal patterns, we constructed a function based on the dynamic time warping algorithm called MatchPattern() that takes in signal readings from specific sensors, such as the accelerometer, sees whether the current series of signals match with a specified pattern, and returns a Boolean value to the user. This allows students to use more powerful constructs in their program, but frees them from having to work with overly complex syntax or algorithms.
More interestingly, as the program is constructed block by block, the source code is generated instantaneously next to the "icon-laying" layer. This gives students an idea of what they code actually looks like. In addition, this code can be cut-and-pasted into a normal text editor for further modification before compiling and downloading to the microcontroller. As described in the Evaluation section, we noticed that many students would take this step on their own initiative and explore the syntax of the program code themselves.
In addition to the tools and platform, the syllabus of any wearable computing workshop requires some thought. The cross-disciplinary nature of wearable computing requires that workshops or courses that utilize it as a learning platform will have to include concepts in electricity and electronics. This is similar to workshops or courses that utilize robotics as a learning platform; even if the primary goal is not to teach about robotics, concepts such as mechanics, motors, and gear ratios usually have to be covered.
Table 1 proposes a syllabus for a 5-day outreach workshop [ 23] for middle-school (Grades 7-9) children without any programming or electronics background. To ensure that participants will acquire the necessary concepts to build their own wearable computing outfit, the course covers concepts from basic electricity and circuit theory to electronics, and finally, to programming. Each of the concepts is reinforced by a set of tasks. The beginning tasks were close-ended and relatively simple. For example, a set task might simply involve asking students to connect LEDs and batteries in series and in parallel, or to create a short circuit and observe the result.
As the workshop progresses, the tasks become more open-ended to encourage creativity from the students. For example, the learning outcomes for complex circuitry require students to know about breadboards, predesignated strips, and jumper wires. To reinforce these learning outcomes, one of the tasks required students to light up four LEDs using as few jumper wires as possible, but the positions of the LEDs were up to the students' imagination.
After electrical theory, students move on to wearable computer programming. As an intermediate step to introduce them to digital logic, they are first issued with a microcontroller chip that has been preprogrammed to cycle through its outputs iteratively. To reinforce the concept of logic signals, students are then asked to design and construct circuits with LEDs flashing in prespecified patterns, or with multicolored LEDs cycling through a series of colors. For instance, one of our tasks required the students to use two multicolored LEDs, placed on the back of the t-shirt, with one flashing first red, then orange, and finally, green; and the other one flashing first aqua, then blue, and finally, purple. After they have grasped the concepts of digital logic, the students learn basic programming concepts such as conditionals, loops, and sequential logic. Tasks for this learning outcome include using light sensors to control LEDs by turning them on or off or flashing different colors depending on the level of ambient light. This requires students to known how to read the analog signals from the light sensors and accelerometers as well as how to output digital signals to control physical devices such as LEDs. For instance, one of our first tasks was to require the students to create a toggle switch using a light sensor. Other tasks involved the control of LEDs using other sensors, such as the accelerometers. For instance, students were asked to use one multicolored LED, which had to turn red when the left arm was moving along the $x$ -axis, turn blue if the motion was along the $y$ -axis, and green when the motion was along the $z$ -axis.
Throughout the entire course, students are encouraged to exercise their creativity and sense of esthetics via open-ended tasks, as well as creating a group project at the end of the workshop. The objective of the group project is to exercise the students' creativity as well as their newly learned programming and wearable computing knowledge.
In contrast with other previous wearable computing workshops [ 14], [ 15], [ 16], our workshop focuses on the learning of electrical, electronic, and programming concepts through wearable computing, while other efforts generally emphasize the construction and design aspect. For example, Buechley et al [ 16] proposes and presents a syllabus in which half of the time is devoted to design and construction of a final deliverable.
In order to validate our platform, we organized a 5-day summer workshop [ 23] with 25 students of ages from 11-16. The students were recruited through open advertisements, and they came from a diverse variety of academic backgrounds. There were 19 boys and 6 girls, who were then divided into nine groups of three students each. The instructors for the workshops were drawn from first-year computer science undergraduates who were teaching the workshop as part of their summer jobs. Compared with other similar workshops [e.g., 16], our participants are younger in age and the class size is larger.
To evaluate whether our platform achieved its requirements, we wished to investigate it along the following dimensions:
Usability of the TeeBoard
Achievement of Learning Outcomes
Comparison with Sewing-based Methods
Being the construction and experimentation platform, we anticipated that the TeeBoard would be subject to a lot of rough treatment from the participants, and that was certainly the case. Our participants, especially the boys, handled the equipment with expected carelessness: connections were ripped from their sockets; batteries were plugged in wherever they would fit, etc. However, even though there were a number of potentially damaging situations, such as short circuits that were created through misconnecting the strips, we are pleased to report that none of the nine TeeBoards failed.
Since our platform was designed to accommodate minimal skill on the part of the user, it is significant that none of the student groups had any major difficulties in completing the close-ended tasks that were designed to reinforce and test their grasp of concepts such as conductivity, connectivity, and logic signals. The programming tasks were also completed satisfactorily, and on several occasions, we noticed students examining and modifying the source code that was generated by BrickLayer. Confirming our intuitions, most of the students employed a trial-and-error approach to solving the tasks: many of them also worked incrementally to complete their circuits, solving one subproblem before moving on to tackle another.
One of the requirements that have to be met before wearable computing can be widely deployed in educational computing is that the associated equipment should be reusable. On average, we estimate that about 10-15 percent of the wires and electronics (in particular, the LED lights, as they have "legs" that break off easily) will need to be replaced after a course series. Since we are working with fabrics and garments that will inevitably be soiled, it is important that the integrity of the strips and connective interface of the TeeBoard be preserved even when it is sent through the wash cycle. To test this out, we subjected one randomly-chosen TeeBoard to a series of increasingly tough wash tests: it was first hand-washed and line-dried, then machine-washed and line-dried, and finally, machine-washed and machine-dried.
The TeeBoard survived all the wash tests without any visible deterioration of its conductive strips or the snap button connective interfaces. In addition, a multimeter test did not detect any deterioration of the conductivity of the conductive traces and interfaces. Finally, the circuit test confirmed that the integrity of the TeeBoard's connective infrastructure was unaffected by the washing and drying.
As one of the attractions of wearable computing is the creativity and esthetics factor, it is important that the teaching platform be able to afford students the space and the support to exercise their creativity.
Figs. 7 and 8 illustrate the results of the open-ended tasks and some of the more outstanding student projects from the summer workshop. Fig. 7 shows a project that uses the accelerometer sensor to control LEDs that blink in particular patterns when the wearer waves his hands. Fig. 8 shows two projects that demonstrate the creativity that the learning platform affords to the students. In one project, a light sensor that is positioned at the front of the shirt controls LEDs that light up when the wearer is walking in the dark (the students called this the "pedestrian taillight"). In the other project, the students created a t-shirt with electronic components and wires that suggest the features of a face. Two white LEDs created the eyes, two multicolored LEDs stood for the cheeks, the microcontroller and the battery pack make up the nose, and a red ribbon forms the mouth. An accelerometer sensor was placed on the "forehead" of the face and programmed so that the multicolored LEDs would "blush" red when the forehead was patted.
Fig. 7. A student demonstrates his project: a t-shirt that blinks when the wearer waves his hands.
Fig. 8. Student projects from our wearable computing workshop. (a) Two boys present their project to the class: a safety t-shirt with light sensors at the front of the shirt that control LEDs at the back. (b) A girl presents the project from her group: a t-shirt with a smiley "face" that "blushes" when the "forehead" is patted.
When the instructors were polled, the general consensus that they expressed was that the easy construction and deconstruction of wearable computing creations afforded by the TeeBoard, together with the easily understood interface of BrickLayer, meant that the students were less intimidated by the topic and the associated technical concepts, and were more willing to try new things on their own, leading to a faster learning process. In addition, the negligible imperfections and constraints of the materials used in the TeeBoard made it possible for the participants to focus on the circuit design and construction without having to take other factors such as resistance and conductivity into mind.
In contrast, a parallel workshop that was held during that same summer that used the conventional approach of sewing on electronic circuits and programming in C experienced substantial trouble getting its creations to work. Most of its problems revolved around making good connections, mainly because it was not easy to create a stable and reliable connection between the electronic components and the conductive thread wires with needle and thread. In addition, the effort expended on constructing the circuits was so substantial that the students were more cautious of making mistakes. As a result, despite the students being more advanced in their studies, the progress of that workshop was significantly slower, and the students' projects were more modest.
Two modes of assessment were used in order to ascertain the knowledge acquired by the students as a result of this camp. First, we used precourse and postcourse surveys in which the students assessed themselves on their knowledge prior to and after going through the workshop. The second method was to use "report cards" that were designed to assess the achievement (or lack thereof) on each learning outcome, for each student group.
Fig. 9 shows the results of the precourse survey, which focuses on the background of the students. All of the participants had an interest in science subjects, but they did not have much prior programming experience or background. Personal conversations and in-class observations further confirmed the fact that even though they may have been exposed to computing before, most of them did not have much knowledge of programming concepts.
Fig. 9. Summary of precourse survey questions and data.
Fig. 10 shows the results of the postcourse survey, which was held to ascertain the students' acquisition of concepts taught during the course, as well as to gauge their interest in the subject matter. It can be seen that students do indeed feel that they have acquired more technical knowledge from the course—in particular, over 50 percent of them reported that they had learned more about programming. This is in contrast to earlier work [ 22], in which it was reported that one of the problems of wearable computing as an educational domain was that students spent too much time sewing and not enough on programming. We believe that this can be attributed both to the easy construction afforded by the learning platform, as well as the beginner-friendly interface of the programming environment.
Fig. 10. Postcourse survey question: what did you learn from this course?
Table 2a shows the report card that was used to assess the achievement (or lack thereof) of each learning outcome per student group, graded for a typical group. It is notable that none of the eight groups had any trouble completing the set tasks, or to create a final project. Most of their final project products were very creative and showed that the students had spent time thinking about how to make their design both creative and technologically challenging. On the report card, all of the eight groups were able to attain at least a "mostly achieved" on all of the learning outcomes; indeed, we were able to assess most of the groups as having completely achieved their learning outcomes.
Our final dimension of evaluation is in the effectiveness of wearable computing as a domain to stimulate the interest of the participants in science and computational subjects. This can be ascertained from the postcourse survey, the results of which are shown in Figs. 11 and 12. From the survey, our workshop is at least partially successful in motivating them to learn more about science and computing. All of the students found the workshop topics interesting and wanted to learn more about science or computational subjects. More than half of the students felt that they would take science subjects based on their experience in this class. This result is gratifying as it shows our workshop is successful in encouraging students to learn science or computing.
Fig. 11. Postcourse survey question: which part of the course did you enjoy the most?
Fig. 12. Postcourse survey questions and data.
Part of the survey also aimed to investigate the level of difficulty of our workshop syllabus. Among the participants, one boy found the course easy, while the others found it difficult. This result is to be expected, as wearable computing would be new and unfamiliar to the students. In addition, most of the students felt that the programming was difficult. This is due to the nature of programming the microcontroller, as it involves physical constraints as well as logical issues. As an example, when asked to program a toggle switch to control the LED, most students were able to come up with the idea of creating a variable to store the state of the LED (on or off) by themselves. However, when asked to write a program to flash the LED on and off, most of the students did not realize that a timed delay was necessary to slow down the program so that the LED's light would be visible to the human eye!
Overall, students gave very high ratings to our workshop. They especially felt that our course was interesting in the design, circuitry, and programming sections. Therefore, we believe that wearable computing is a suitable topic through which we can raise the level of exposure and awareness of middle school children toward technological and computing issues.
In contrast to previous wearable computing workshops [ 15], which used subject experts such as researchers in the field, our instructors were all first-year undergraduate students teaching this course as part of their summer jobs. This was done to test whether it would be possible to run outreach workshops in wearable computing even if highly trained instructors were not available.
In general, our instructors did not report much difficulty in teaching or demonstrating the course material. When asked to comment on the students' performances, they reported that while the students did not show much interest in learning about electrical theory (voltage, conductivity, resistances, etc), they became excited and animated when they were presented with the TeeBoard and the BrickLayer platforms and immediately started trying to make their own constructions. They experimented on their own, connecting the LEDs and wires in all sorts of configurations and even trying color mixing on their own. This confirms our intuition that wearable computing is an appropriate teaching medium as it is interesting and exciting to the students.
In performing the assigned tasks, our instructors observed that most of the all-male groups focused on finishing the tasks as quickly as possible, without paying much attention to the esthetics of the final product. In contrast, the all-female groups worked more slowly and methodically, considering factors such as design, color, and patterns. They also inserted their own interpretations into their assignments: for instance, one group designed a circuit and programmed the color of the multicolored LED to change according to the colors of the rainbow. When asked to explain their idea, they stated that they chose the rainbow pattern as it represented hope.
According to the instructors, the students experienced the most difficulty when first presented with the TeeBoard, as they did not really understand the whole structure of the conductive strip pattern at first. Most of them constructed their circuits using direct connecting wires rather than exploiting the strips on the TeeBoard. However, once the concept was explained to them, they picked it up quickly and had no trouble using the TeeBoard in the way that it was designed for. The second most apparent difficulty was that of the preprogrammed microcontroller, as many of them did not understand that the power terminals of the chip had to be connected in a specific manner. They also had some trouble understanding the concept of a preprogrammed chip. However, these problems were easily overcome with some additional explanation and demonstration.
For comparison, we also organized another wearable computing workshop, which used "traditional" methods such as sewing and C programming. This workshop was similarly held over a period of five full days, and attended by six students of ages 17-19, who were divided into two groups and taught by three wearable computing experts. The learning outcomes for this workshop are shown in Table 2b. Compared with the participants in the workshop that utilized our learning platform, these students were older and more academically advanced, and therefore, some of the learning outcomes (such as electric circuitry) was not applicable for them.
For the programming learning outcome, some of the students in the conventional workshop had prior programming background and knew the C language. Those students tended to "take charge" of the programming parts of the wearable computing design. In addition, the usual text-based development environment did not lend itself easily to collaborative design or learning. As a result, those students who already knew programming usually handled the entirety of the programming part; the other students were mostly nonparticipative. This is in contrast to the situation encountered in the other workshop that was using the BrickLayer environment: since there was minimal syntax involved, and the platform was novel to all, the entire group could contribute to the logic design of the program. We believe that this illustrates the need for an environment such as BrickLayer to facilitate the learning of programming.
For the innovative design outcome, we noticed that students in the conventional workshop worked more slowly and methodically on their designs. For example, in contrast to the other workshop that was using the TeeBoard and BrickLayer, they would attempt to come up with a complete plan on paper before starting the construction on their garment. We also noticed that they had a lot of difficulty and frustration when constructing the garment, which we attribute to the difficulty of the sewing approach.
Partially as a result of the difficulties that they encountered, the creations from the conventional workshop participants were more conservative and less imaginative. For example, they tended to copy from the designs and the ideas that we showed them as examples, as opposed to the participants from the other group, who were more willing to make mistakes and try out new things on their own. In addition, they also tended not to iterate over their constructions—their first successful construction was the final one—in contrast to the participants from the other workshop, who by and large tended to iteratively improve over their construction, ripping out and rebuilding parts of it, even though it was working functionally. We believe that this illustrates the ability of the TeeBoard and the BrickLayer platforms to effectively support the learning of innovative technologies as well as to encourage creativity.
We have presented a learning platform for wearable computing that comprises of a user-friendly construction platform, a hybrid text-graphical programming environment, and a sample syllabus that guides students through the basic concepts involved in wearable computing.
The proposed platform was evaluated through deploying it in a summer outreach workshop and investigating its performance along eight dimensions. The results and findings show that our platform fulfills its requirements in being a feasible and effective learning environment. Our comparative evaluation with a traditional sewing-based wearable computing workshop also shows that the Teeboard and Bricklayer-based approaches are more successful at achieving certain learning outcomes, such as complex circuitry, wearable computing programming, and innovative design.
Even though our syllabus was designed for an outreach workshop and our evaluation was carried out through deployment in such a workshop, it does not limit the use of wearable computing to such workshops. Indeed, there is no reason why such a platform cannot be used in higher-level courses, such as college-level introductory programming. Obviously, such a course would not require all the components of the platform that we have described here: it would likely have a very different syllabus and would perhaps forgo the BrickLayer environment in favor of using conventional textual programming, but the TeeBoard construction platform still fulfills its goals of freeing students from the constraints and considerations of material imperfections and the time-consuming, low-level work of sewing and soldering, and allows them to focus on the higher-level task of programming and design. In other words, the platform still fulfills its goals of making it feasible to integrate wearable computing into education.
In the future, we plan to deploy the learning platform in larger-scale workshops as well as in regular curriculum courses, both at the high school and at the university level. We also plan to further evaluate our platform via quantitative measurements and comparisons with traditional programming and electronics education, as well as similar innovative workshops based on robotics.
This work was partially supported in part by two grants (project codes 8CGF and G-U529) from the Hong Kong Polytechnic University and the eToy Laboratory of the Department of Computing.