Bayesian Probability-Based Motion Estimation Method in Ubiquitous Computing Environments

Abstract

Natural User Interface/Natural User Experience (NUI/NUX) is one of the core techniques to control deployed devices in ubiquitous computing environments. User friendly user interfaces can be provided by estimating and utilizing locations and gestures/postures of users. However, the recognition of users’ gestures/postures is limited because of the number of the available deployed motion recognition sensors. This paper proposes a Bayesian probability-based motion estimation method to control users’ unestimated motions. Given that the whole postures/gestures can be predicted based on the estimated motion, the intention of user can be also predicted. In the experiments, Myos were utilized as motion recognition sensors. The proposed method was validated showing the result through a virtual character. By collecting the data of users’ motions in advance, unmeasured motions could be deducted by the proposed method.