Hiroyuki Manabe

Full-time wearable headphone-type gaze detector

A headphone-type gaze detector for a full-time wearable interface is proposed. It uses a Kalman filter to analyze multiple channels of EOG signals measured at the locations of headphone cushions to estimate gaze direction. Evaluations show that the average estimation error is 4.4® (horizontal) and 8.3® (vertical), and that the drift is suppressed to the same level as in ordinary EOG. The method is especially robust against signal anomalies. Selecting a real object from among many surrounding ones is one possible application of this headphone gaze detector.

Multi-stream HMM for EMG-based speech recognition

A technique for improving the recognition accuracy of EMG-based speech recognition by applying existing speech recognition technologies is proposed. The authors have proposed an EMG-based speech recognition system that requires only mouth movements, voice need not be generated. A multi-stream HMM (hidden Markov model) and feature extraction technique are applied to EMG-based speech recognition. 3 channel facial EMG signals are collected from ten subjects when uttering 10 Japanese isolated digits. One channel corresponds to one stream. By examining various features, we found that the delta component of the static parameter leads to higher accuracy. Compared to equal stream weighting, the individual optimization of stream weights increased recognition accuracy by 4.0% which corresponds to a 12.8% reduction in error rate. This result shows that multistream HMM is effective for the classification of EMG.

Direct Gaze Estimation Based on Nonlinearity of EOG

Electrooculography (EOG) is one of the measures used to estimate the direction of a persons gaze; however, conventional EOG techniques suffer from a drift issue which makes it difficult to extract an accurate absolute eye angle. The technique proposed here is based on the nonlinearity of the EOG and offers a practical solution to this problem. It estimates the absolute eye angles before and after a saccade, which cancels the offset due to the drift. Additionally, it does not require any effort from the user or any target, but instead uses only the difference of the EOGs. Experiments with five subjects confirm that the proposed technique can estimate the absolute eye angle with an error of less than

Single capacitive touch sensor that detects multi-touch gestures

4^\circ

A ring-shaped EMG measurement system for applying to user interface

. They also show improvements are achieved with several options such as weighting and multiple saccades. The technique will contribute to practical EOG-based interaction systems.

Eye Tracking Using Built-in Camera for Smartphone-based HMD

A technique that allows a single capacitive touch sensor to recognize multi-touch gestures is proposed. Touch, multi-finger swipe and swipe direction are recognized. It does not need a multiplexer or complicated wiring and well suits wearable devices. An experiment with 8 subjects confirms that the proposed technique can recognize multi-touch gestures.

Tag system with low-powered tag and depth sensing camera

This paper proposes a ring-shaped EMG measurement system to be worn on one or more fingers for a new user interface that relies on EMG measurements. A preamplifier is mounted on each electrode to suppress the effect of noise. The system is truly wearable since the main amplifier is designed to be worn on the wrist. The proposed EMG measurement system is implemented and evaluated. The experiments confirm that the proposed electrode is superior to conventional electrodes in terms of stability and usability while offering excellent long term performance. The effectiveness of the proposed electrode system is confirmed by the test in which the interface realizes highly accurate speech recognition via EMG.

Headphone taps: a simple technique to add input function to regular headphones

Virtual reality (VR) using head-mounted displays (HMDs) is becoming popular. Smartphone-based HMDs (SbHMDs) are so low cost that users can easily experience VR. Unfortunately, their input modality is quite limited. We propose a real-time eye tracking technique that uses the built-in front facing camera to capture the users eye. It realizes stand-alone pointing functionality without any additional device.

AwareLESS authentication: insensible input based authentication

A tag system is proposed that offers a practical approach to ubiquitous computing. It provides small and low-power tags that are easy to distribute; does not need a special device to read the tags (in the future), thus enabling their use anytime, anywhere; and has a wide reading range in angle and distance that extends the design space of tag-based applications. The tag consists of a kind of liquid crystal (LC) and a retroreflector, and it sends its ID by switching the LC. A depth sensing camera that emits infrared (IR) is used as the tag reader; we assume that it will be part of the users everyday devices, such as a smartphone. Experiments were conducted to confirm its potential, and a regular IR camera was also tested for comparison. The results show that the tag system has a wide readable range in terms of both distance (up to 8m) and viewing angle offset. Several applications were also developed to explore the design space. Finally, limitations of the current setup and possible improvements are discussed.

Robust and Preceding Speech Detection Using EMG

A simple technique which changes regular headphones into input-and-output devices is proposed. It detects headphone taps and also captures users voice. Two prototypes are implemented. Users can control music players and have phone conversation via their favorite headphones without attaching external switches or microphones. We confirm that they work well with many headphones in various environments.