Showing papers presented at "International Symposium on Wearable Computers in 2006"

Discovering Characteristic Actions from On-Body Sensor Data

Abstract: We present an approach to activity discovery, the unsupervised identification and modeling of human actions embedded in a larger sensor stream. Activity discovery can be seen as the inverse of the activity recognition problem. Rather than learn models from hand-labeled sequences, we attempt to discover motifs, sets of similar subsequences within the raw sensor stream, without the benefit of labels or manual segmentation. These motifs are statistically unlikely and thus typically correspond to important or characteristic actions within the activity. The problem of activity discovery differs from typical motif discovery, such as locating protein binding sites, because of the nature of time series data representing human activity. For example, in activity data, motifs will tend to be sparsely distributed, vary in length, and may only exhibit intra-motif similarity after appropriate time warping. In this paper, we motivate the activity discovery problem and present our approach for efficient discovery of meaningful actions from sensor data representing human activity. We empirically evaluate the approach on an exercise data set captured by a wrist-mounted, three-axis inertial sensor. Our algorithm successfully discovers motifs that correspond to the real exercises with a recall rate of 96.3% and overall accuracy of 86.7% over six exercises and 864 occurrences.

A Construction Kit for Electronic Textiles

Abstract: Construction kits have long been popular as educational artifacts, supporting and encouraging creative explorations of engineering and design; but to date, such kits have had little connection with the new and expanding field of electronic textiles (e-textiles). We believe that creating an "e-textile construction kit" could provide a powerful new medium to engage a diverse range of students in electrical engineering and computer science. This paper, then, describes a construction kit designed to introduce novices to electronics, computing and design via e-textiles. We describe each component of the kit, provide examples of constructions that were built with the kit, and examine the durability of these constructions. We conclude with a discussion of the results of preliminary user testing and an exploration of our plans for continued work in this area.

Textile Pressure Sensor for Muscle Activity and Motion Detection

Abstract: This paper presents a pure textile, capacitive pressure sensor designed for integration into clothing to measure pressure on human body. The applications fields cover all domains where a soft and bendable sensor with a high local resolution is needed, e.g. in rehabilitation, pressure-sore prevention or motion detection due to muscle activities. We developed several textile sensors with spatial resolution of 2 times 2 cm and an average error below 4 percent within the measurement range 0 to 10 N/cm2. Applied on the upper arm the textile pressure sensor determines the deflection of the forearm between 0 and 135 degrees due to the muscle bending.

Augmenting spatial awareness with Haptic Radar

Abstract: We are developing a modular electronic device to allow users to perceive and respond simultaneously to multiple spatial information sources using haptic stimulus. Each module of this wearable "haptic radar" acts as an artificial hair capable of sensing obstacles, measuring their range and transducing this information as a vibro-tactile cue on the skin directly beneath the module. Our first prototype (a headband) provides the wearer with 360 degrees of spatial awareness thanks to invisible, insect-like antennas. During a proof-of-principle experiment, a significant proportion (87%, p = 1.26 * 10-5) of participants moved to avoid an unseen object approaching from behind without any previous training. Participants reported the system as more of a help, easy, and intuitive. Among the numerous potential applications of this interface are electronic travel aids and visual prosthetics for the blind, augmentation of spatial awareness in hazardous working environments, as well as enhanced obstacle awareness for motorcycle or car drivers (in this case the sensors may cover the surface of the car).

Combining Motion Sensors and Ultrasonic Hands Tracking for Continuous Activity Recognition in a Maintenance Scenario

Abstract: We present a novel method for continuous activity recognition based on ultrasonic hand tracking and motion sensors attached to the user's arms. It builds on previous work in which we have shown such a sensor combination to be effective for isolated recognition in manually segmented data. We describe the hand tracking based segmentation, show how classification is done on both the ultrasonic and the motion data and discuss different classifier fusion methods. The performance of our method is investigated in a large scale experiment in which typical bicycle repair actions are performed by 6 different subjects. The experiment contains a test set with 1008 activities from 21 classes encompassing 115 minutes randomly mixed with 252 minutes of 'NULL' class. To come as close as possible to a real life continuous scenario we have ensured a diverse and complex 'NULL' class, diverse and often similar activities, inter person training/testing and an additional data set only for training (299 extra minutes of data). A key result of the paper is that our method can handle the user independent testing (testing on users that were not seen in training) nearly as well as the user dependent case.

Towards Less Supervision in Activity Recognition from Wearable Sensors

Abstract: Activity Recognition has gained a lot of interest in recent years due to its potential and usefulness for context-aware wearable computing. However, most approaches for activity recognition rely on supervised learning techniques lim iting their applicability in real-world scenarios and their scalability to large amounts of activities and training data. State-of-the-art activity recognition algorithms can roughly be divided in two groups concerning the choice of the classifier, one group using generative models and the other discriminative approaches. This paper presents a method for activity recognition which combines a generative model with a discriminative classifier in an integrated approach. The generative part of the algorithm allows to extract and learn structure in activity data without any labeling or supervision. The discriminant part then uses a small but labeled subset of the training data to train a discriminant classifier. In experiments we show that this scheme enables to attain high recognition rates even though only a subset of the training data is used for training. Also the tradeoff between labeling effort and recognition performance is analyzed and discussed.

Trinetra: Assistive Technologies for Grocery Shopping for the Blind

Abstract: Trinetra aims for cost-effective, assistive technologies to provide an independent grocery-shopping experience for the blind by leveraging barcodes and networking diverse embedded COTS devices.

Design and Evaluation of a Wearable Optical Sensor for Monitoring Seated Spinal Posture

Abstract: This work describes the development and evaluation of a wearable plastic optical fiber (POF) sensor for monitoring seated spinal posture. A garment-integrated POF sensor was developed and tested on nine healthy subjects, and its performance compared to data taken simultaneously from a marker-based motion capture system. Sensor performance correlated strongly with motion-capture data with an average r2 of 0.913. Results show that the wearable sensor provides enough accuracy of measurement to reliably monitor seated spinal posture.

Assessing the Wearability of Wearable Computers

Abstract: In terms of mounting a computer on the body, the computer's weight, size, shape, placement and method of attachment can elicit a number of effects. Inappropriate design may mean that the wearer is unable to perform specific tasks or achieve goals. Excessive stress on the body may result in perceptions of discomfort, which may in turn affect task performance, but ultimately raises issues of health and safety. This paper proposes a methodology for assessing the affects of wearing a computer in terms of physiological energy expenditure, the biomechanical effects due to changes in movement patterns, posture and perceptions of localised pain and discomfort due to musculoskeletal loading, and perceptions of well- being through comfort assessment. From ratings of these effects the paper proposes 5 levels to determine the wearability of a computer.

Workers' Routine Activity Recognition using Body Movements and Location Information

Abstract: We propose a method for recognizing workers' routine activities by combining location information and body movements of a user. We describe methods for capturing and recognizing the nursing context with an infrared-ID location sensor system and accelerometers worn by a user. We show experimental results of recognizing typical activities in a nursing scenario in a laboratory setting.

VibeFones: Socially Aware Mobile Phones

Abstract: Today's mobile phones are essentially off-the-shelf, mass-market wearable computers. In this paper, we describe mobile social software that uses tone of voice, location and proximity information to create a sophisticated understanding of people's social lives, by automatically mining their face-to-face and phone interactions. We describe several applications of our system - automatic characterization of social and workplace interactions, a courtesy reminder for phone conversations, and a personal trainer for dating encounters.

ActionGSR: A Combination Galvanic Skin Response-Accelerometer for Physiological Measurements in Active Environments

Abstract: The galvanic skin response (GSR), also known as elec- trodermal response, measures changes in electrical resistance across two regions of the skin. Galvanic skin response can measure arousal levels in children with autism; however, the GSR signal may be overwhelmed by the vigorous movements of the children. This paper introduces ActionGSR, a wireless sensor capable of measuring both GSR and acceleration simultaneously in an attempt to dis ambiguate valid GSR signals from motion artifacts.

Wearable Band Using a Fabric-Based Sensor for Exercise ECG Monitoring

Abstract: Recently, smart textiles, especially for wearable garments that capable of recording vital signals, have attracted a lot of attention. There are many reports about wearable garment-based platforms focusing on the monitoring of ECG, but most of them are for static situation. In this paper, we report a fabric-based garment sensing system to monitor the exercise electrocardiogram (ECG); a wireless communication configuration based on bluetooth to transfer the signals into a PDA; and an algorithm to retrieve the desired ECG signals from the raw data. In addition, PDA can communicate with a remote server by WiFi to insure that all data could be stored exactly. The experimental results in resting, walking, or jogging, show this wearable garment system can steadily, continuously and precisely get the ECG, RRI and HRV.

Low Power Wearable Audio Player Using Human Body Communications

Abstract: This paper presents a prototype wearable audio player system to playback the digital audio signal transmitted through the wearer's body without any wire. To significantly reduce the power consumption of the digital audio transmission, the system exploits a novel wideband signaling human body communication scheme. The transceiver chip designed, fabricated, and assembled on the prototype system shows 5-mW power consumption at 2-Mb/s data rate, better than the Bluetooth transceiver. Real-time audio playing through the human body is successfully demonstrated.

A Wearable Textile for Monitoring Respiration, Using a Yarn-Based Sensor

Abstract: In the last few years, smart textiles with fabric-based sensors to monitor gesture, posture or respiration have become popular. Although the fabrics seem to have nice performance to some extent, the 2-D form of fabric-based sensors limits the resolution. In this paper, a yarn-based sensor is developed. The performance of the sensors relies on the structure of the yarns and the response while extended depends on the twists per meter (TPM) of the yarns. For verifying the feasibility, a simulation system based on respiration monitor using yarn-based sensors is reported. We find the sensors output can closely track the periodic input signals.

Wearable Sensor Network for Body Kinematics Monitoring

Abstract: A comfortable wearable sensor network for monitoring the body kinematics during hidrocinesiotherapy sessions is presented. Sensing modules composed by 3-axis accelerometers, 3-axis magnetometers and interface electronics are integrated in a swimming suit and connected to a microcontroller by a serial interface. Data from the suit is transmitted to a base station using a 2.4 GHz RF transceiver. Temperature, heart rate and frequency rate are also monitored by the suit, and the sensor network presented can distinguish between rotation and motion of the monitored system. Each sensing module presents a 1 degree resolution.

User Localization Using Wearable Electromagnetic Tracker and Orientation Sensor

Abstract: This paper describes a localization method with wearable electromagnetic sensor and orientation sensor for wearable computer users. Many user localization methods have been investigated to realize location-based services in a wide environment. The localization methods usually employ a hybrid approach in which user's position is estimated by using positioning infrastructures and dead reckoning such as a pedometer. However, the installation cost of infrastructures increases when the area expands, and the error of the pedometer is frequently caused by failures in walking locomotion detection and the difference between ideal and estimated step lengths. If the relative distance is accurately estimated by dead reckoning approach, the installation cost of infrastructures can be reduced. This paper proposes a new localization method that improves the estimation accuracy of step length in dead reckoning approach. The proposed method measures user's orientation and geometrical relationship between user's heel and waist with an orientation sensor and an electromagnetic tracker attached to user's body. When both feet come into contact with the ground, user's position is updated by adding the estimated step length which means the relationship between user's heels and the position estimated in the previous step. The proposed localization method has been experimented using a prototype system to evaluate the accuracy of the proposed method.

A Low-power Star-topology Body Area Network Controller for Periodic Data Monitoring Around and Inside the Human Body

Abstract: More than hundred sensor devices are required to monitor various body signals. The body area network (BAN) connecting the sensor devices with the controller needs low power consumption and real-time operations, which is difficult to implement by conventional controller chips. In this paper, a low power controller chip is designed and fabricated to manage the proposed network with low power consumption. It periodically monitors the data from maximum 255 sensor devices with less than 2-mW power consumption. The controller is implemented in 25 mm silicon area and its operation is successfully demonstrated on the test system board.

Managing Smart Garments

Abstract: Ease of maintenance and management of smart garments (garments with integrated electronics) is crucial to their user acceptability and commercial viability. This paper presents a system that addresses user needs of easy garment charging, storage, and synchronization. The HBar smart hanger system is a novel system consisting of a set of augmented garments and coat hangers to fulfil the requirements of a smart garment management system.

Evaluation of Four Wearable Computer Pointing Devices for Drag and Drop Tasks when Stationary and Walking

Abstract: This paper presents the results of two experiments comparing four commercially available pointing devices performing drag and drop tasks for use with wearable computers. The pointing devices evaluated consist of a Trackball, Touchpad, Gyroscopic mouse and Twiddler2 mouse. The studies involved 24 participants performing drag and drop tasks with the pointing devices while wearing a wearable computer on their back and using a head-mounted display. The first experiment required participants to perform drag and drop tasks when stationary, while the second experiment required participants to perform drag and drop tasks while walking. The error rate and time to complete the drag and drop of the targets was measured and participants were required to complete an exit questionnaire. The results are presented in this paper.

Wearable Biomedical Monitoring System Using TextileNet

Abstract: We developed and tested a biomedical monitoring system using TextileNet, a flexible conductive garment for wearable computing. TextileNet detects biological signals while simplifying communication and power supply wiring. TextileNet also acts as an electromagnetic interference (EMI) shield, which makes it possible to use simpler amplifiers in the system. Using TextileNet, a huge amount of biological information can be processed simultaneously.

FingerMouse - Architecture of an ASIC-based Mobile Stereovision Smart Camera

Abstract: The FingerMouse is a gumstick-sized (43 mm times 18 mm) visual input system for wearable computers. Worn on the body, the system captures the user's hand and outputs in real-time the coordinates as well as a 1-bit image of the hand segmented from a (moving) background. It integrates cameras and vision processing using a custom designed FingerMouse-ASIC, processing images at 20 GOp/s, using only 78 mW power. This paper describes how the architecture of the FingerMouse outperforms PC-based implementations in efficiency and latency, while operating under strict size and power consumption constraints.

Evaluation of an Eyes-Free Cursorless Numeric Entry System for Wearable Computers

Abstract: We report on the results of a user study to investigate the utility of passive haptics for eyes-free numeric entry. This work targets cursorless user interfaces designed for use with a watch-sized wrist-worn computer. Our study compared three approaches for selecting one of a set of eight numeric parameters and entering its value, both with and without visual feedback. The three selection methods utilized physical buttons alone, buttons with a touch-sensor utilizing passive haptics, and the touch sensor with passive haptics alone. The results show that passive haptics allowed users to perform parameter selection and number entry tasks, with statistically insignificant differences in accuracy and speed when used with and without visual feedback. Furthermore, there was no statistically significant difference in accuracy and speed between the button-based methods and the purely touch-sensor?based approaches.

Eye of the Beholder: Phone-Based Text-Recognition for the Visually-Impaired

Abstract: Blind and visually-impaired people cannot access essential information in the form of written text in our environment (e.g., on restaurant menus, street signs, door labels, product names and instructions, expiration dates). In this paper, we present and evaluate a mobile text-recognition system capable of extracting written information from a wide variety of sources and communicating it on-demand to the user. The user needs no additional hardware except an ordinary, Internet- enabled mobile camera-phone - a device that many visually-impaired individuals already own. This approach fills a gap in assistive technologies for the visually- impaired because it makes users aware of textual information not available to them through any other means.

Practical Context Awareness for GSM Cell Phones

Abstract: The ability to provide context aware behaviour on a cell phone such as whether a user is walking or driving has previously only been possible via the use of additional hardware sensors such as an accelerometer. In this paper we show that by using GSM data such as signal strength fluctuation and changes to the monitored neighbouring cells it is possible to distinguish between various states of movement such as walking, driving in a motor car and remaining stationary. We demonstrate how the calibration of the cell phone for use in a given environment can be implemented in an automatic and unsupervised manner, and that we can achieve a classification accuracy of around 80%.

Modelling and Handling Seams in Wide-Area Sensor Networks

Abstract: The fields of wearable computing, augmented reality and ubiquitous computing are in principle highly convergent, as they all promise a Utopian future in which the devices embedded in the environment, our bodies and our clothes will have reached a level integration such that we can intuitively perceive and interact with our environment. However, the reality as practised in research labs and limited commercial deployments has been that budgetary and technical constraints have actually kept these fields separate and distinct. One manifestation of this separation is in the choice of sensors used to build systems in each domain. A truly cross-disciplinary project has to incorporate sensors of much greater heterogeneity than has occurred heretofore. The way in which sensors are deployed results in spatial seams that can act as obstacles to the provision of services across different areas. This paper takes an architectural approach to handling events from different tracking systems and maintaining a consistent spatial model of people and objects. The principal distinguishing feature is the automatic derivation of dataflow network of distributed sensors, dynamically and at run-time, based on requirements expressed by clients.

Designing Wearables

Abstract: One of the perennial criticisms of work in the field of wearable technology is that nobody seems to be wearing it! This is despite stunning advances in the underlying technology. At the Ninth ISWC there was a feeling that wearable computing development needs to progress on the issue of user acceptability. This poster describes a pilot survey of designers working within the field. The results provide a snapshot of wearable technology design practice and how user centered design (UCD) is perceived with the field.

Using the "HotWire" to Study Interruptions in Wearable Computing Primary Tasks

Abstract: As users of wearable computers are often involved in real-world tasks of critical nature, the management and handling of interruptions is crucial for efficient interaction and task performance. We present a study about the impact that different methods for interruption have on those users, to determine how interruptions should be handled. The study is performed using an apparatus called "HotWire" for simulating primary tasks in a laboratory experiment, while retaining the properties of wearable computers being used in mobile, physical, and practical tasks.

Palm-sized Attraction Force Display Exploiting the Nonlinearity of Perception

Abstract: In this paper, we discuss a novel force perception method for mobile and wearable displays. The method exploits the nonlinearity of human perception to generate a force sensation. In our method, brief intense pulses of acceleration alternate with longer periods of low- amplitude recovery. This asymmetric acceleration generates a unidirectional force sensation using a crank- slider mechanism that physically generates bi-directional force. We designed a palm-sized prototype and verified the feasibility of the proposed method through experiments.

Maximum Power Point Tracking (MPPT) for on-body Context Systems

Abstract: In this article we present the use of MPPT (maximum- power-point-tracking) for miniaturized on-body activity recognition systems. We evaluate different solar cells and apply MPPT to study the suitability for on-line activity systems. With a wrist worn solar cell, MPPT allows us to track power levels of 1.35 mW for indoors/wood-workshop lighting and 7.21 mW during dull outdoor lighting conditions. The overall efficiency of the proposed technique is 96% in the best-case scenario and 80% in the worst-case.