Biometric Data Collection Device for User Research

Transcription

1 Biometric Data Collection Device for User Research Design Team Daniel Dewey, Dillon Roberts, Connie Sundjojo, Ian Theilacker, Alex Gilbert Design Advisor Prof. Mark Sivak Abstract Quantitative video game research is cost-restrictive and smaller, independent gaming studios are often unable to perform quantitative user experience studies prior to their game releases. The device described here will enhance user experience testing by providing a simple and integrated, low-cost solution. The device aims to be minimally intrusive during testing while providing high-volume, filtered data from multiple biometric sensors. This data can then be used to contextualize the user s input for analysis of the game s effectiveness. Academic journals and experts within the field of psychophysiology were consulted to understand what biometric inputs correlate to emotional response. Certain aspects of human emotion can be quantified by gathering galvanic skin response, heart rate, grip force, and skin temperature, which can then be compared to the user s self-reported emotional states. Research was completed to determine the current technology of biosensors and how the signals are obtained. After analyzing common sensor placements through experimental means, a device on the hand and wrist was determined to be the best option to provide accurate data while maintaining comfort for the user. By combining all of these sensors into an unobtrusive, affordable device this product will be readily usable by independent game studios and researchers to perform quantitative user experience testing. For more information, please contact m.sivak@husky.neu.edu. 84

2 The Need for Project The video game industry lacks Videogame development, a multibillion dollar industry, currently uses quantitative user experience questionnaire based user experience testing to determine the testing that is not costrestrictive and provides a explanation and training for both the user and facilitator of the test. The effectiveness of their games. These studies often require extensive mobile testing solution. studies are also expensive and require a controlled space. On average a focus group consisting of 8 people costs $6,000. This consists of a 90 minute session with 10 minutes of the moderator talking. Other options are personal interviews which cost $325 per person, 200-person phone surveys which can vary between $5000 to $15000 and mail surveys which cost $5000 to $7000. All these prices depend on the length of the questionnaire (Ref. 1). Furthermore, the questionnaires provided must be worded in a way that prompts definitive answers while not soliciting responses that are bias in any way. All of these requirements lead to a costly testing procedure that only provides qualitative data. User experience testing should be easily automated, mobile, and provide quantitative results. Currently, a simple, reliable, affordable, and wearable device that can capture a user s emotional state does not exist. The Design Project Objectives and Requirements Develop a product that is affordable, reusable, lightweight, unobtrusive to gameplay, and provides accurate quantitative data for user experience testing. Design Objectives For this product to be an effective tool for user experience testing it must provide researchers with accurate and consistent data while not interfering with the gameplay experience. The device must be easy to set up and maintain, requiring little to no subject preparation. Design Requirements The device must be powered for a full day of testing, up to 8 hours, using common batteries. The device should be wireless, utilizing Bluetooth technology. To maintain an unobtrusive testing experience, the device must be light weight, less than 1 pound, and have an adjustable fit. This fit should span the forearm circumference of the 5 th percentile of women, 9.33 inches, and 95 th percentile of men, inches (Ref. 30). To provide a low-cost solution the prototype should cost no more than $

3 Design Concepts Considered Four device location concepts In determining the best location for a biometric device, common were evaluated: wrist, hand, locations on the hands, wrist, and chest were tested. These locations chest, and a wrist & hand were chosen based on existing technology and published research: heart combination rate monitors are commonly placed on the chest; electromyography (EMG) and acoustic myography (AMG) sensors are often placed on the wrist; the fingers have been shown to produce strong signals for infrared (IR) heart rate (HR); and galvanic skin response (GSR) can be measured on the fingers, hand, and wrist (Ref. 2, 3, & 27). The chest placement was eliminated due to lack of sensor functionality. Hand Concept The hand location concept, Figure 1, includes a force sensitive resistor (FSR) placed on the hand by the thumb to monitor changes in the subject s stress as a result of varying grip strength. Electrodes placed on the palm for GSR are sensitive to changes in the subject s sweat (Rep ). A pulse sensor is located on the pinky to remain unobtrusive. As previously mentioned, IR HR data collected from the tips of fingers is found to be accurate, and is commonly used in medical devices (Rep ). Because of the subject s hand motion required to Figure 9: Hand Concept play a game, contact on the hand is not likely to be consistent, restricting the ability of both the GSR and pressure sensor to record reliable data. Moreover, housing for the battery pack, microcontroller, and Bluetooth would be very cumbersome anywhere on the hand. Wrist Concept The wrist location concept, Figure 2, places GSR electrodes on the bottom of the wrist. An IR HR sensor can be placed on the wrist, but existing technology has shown unreliable results. AMG would be Figure 2: Wrist Concept placed on the top of the wrist to measure muscle activity from thumb movement. The battery casing would also be on the top of the wrist. Hand & Wrist Combination Concept The hand and wrist concept, Figure 3, combines the strong IR HR signal of the finger and the motion-resistant high-contact areas of the wrist for GSR. In this concept, the IR HR sensor remains unobtrusive on the pinky and GSR remains in contact, unobtrusive on the wrist. The pressure sensor can detect grip changes on the hand, and the housing Figure 3: Hand & Wrist becomes less cumbersome on the back of the wrist. Combination Concept 86

4 Recommended Design Concept Using a combination of wrist Design Description and hand sensors is the best The hand and wrist combination device consists of two way to collect accurate data in components: a glove and a plastic housing, Figure 4. The glove an unobtrusive way. contains the GSR, pulse, and pressure sensors and has straps on the wrist to be tightened to ensure a strong connection. The GSR electrodes are located on the front of the wrist, the pressure sensor on the palm, and pulse sensor on the pinky. The plastic housing contains the microcontroller, Bluetooth, power supply, accelerometer, and temperature sensor. The housing is attached to the forearm using two straps. A hole in the housing allows for the routing of wires out of the enclosure and a snap connector between the glove and enclosure is used for easy removal. Experimental Investigations GSR was tested on the wrist, palm, and fingers to determine the Figure 4: Plastic Housing best location for measurement. A tester first played a low-stress game for 5 minutes and then a high stress game for 5 minutes. It was found that the GSR mean value on the fingers increased by 466 mv with a standard deviation increase from 130 mv to 140mV when aroused. By comparison palm values increased by 700 mv with standard deviation increasing from 279 mv to 450 mv and wrist mean value increased by 138 mv with standard deviation increasing from 15 mv to 112 mv. Although the wrist was the least sensitive position the low standard deviation in the non-aroused state indicates that the stable contact of the wrist can outweigh its sensitivity. Force resistive sensors were tested to determine which location of the hand in contact with a controller produces the best results: the ring finger, upper thumb muscle, or lower thumb muscle. The same procedure as the GSR tests was used. It was found that the lower palm mean increased by 1100 mv and the standard deviation increased from 172 mv to 350 mv, the upper palm mean increased by 950 mv with a standard deviation decrease of 18 mv, and the ring finger mean decreased by 100 mv and standard deviation increased from 141 mv to 352 mv. Based on these results the lower palm is the best position for the force resistive sensor because the increase in standard deviation and mean value are vindictive of a tighter grip with more variation. 87

5 Key Advantages of Recommended Concept The hand and wrist combination device combines the best possible sensor locations for accurate data collection while avoiding spots prone to contact issues. The design has flexibility since sensor locations can be tightened or loosened depending on user hand and wrist size. This device does not require any disposable parts or skin preparation making it simple to set up and use on a regular basis. Financial Issues The prototype cost is USD. A primary goal for the device is to be affordable for small game studios and independent developers. With this idea in mind the parts used in this device were kept inexpensive. The prototype cost USD in total with the sensors and microcontroller varying between 10 to 25 USD. With bulk purchasing of components and improvements in manufacturing the cost can be reduced at least 50 percent. Recommended Improvements Some recommended improvements include better data processing, implementing a pulse sensor on the wrist and AMG on the forearm, and compressing the electronics into one printed circuit board. One way to improve the device would be to implement a pulse sensor on the wrist. This would make the device less intrusive to gameplay. Adding AMG on the forearm would also reduce intrusiveness and mitigate noise due to unstable contact. Another improvement would be to place all the electronics into one centralized printed circuit board. This would reduce the size of the casing and simplify the design. 88