単眼カメラを用いた動物体対応SLAM(<特集>3次元インタラクション)

抄録

The augmented reality (AR) technology is based on environment recognition systems which can estimate 3D environment in real-time. In this paper, we introduce a new environment recognition system which can estimate not only 3D environment consisted of static feature-points but also poses of shape-and-pattern-known dynamic objects with a single camera simultaneously. Our system is based on the Extended Kalman Filter (EKF). The EKF in our system has state variables which have camera pose, static feature-point positions and dynamic object poses. As the EKF updates the state variables with measurement data by camera, our system can estimate the dynamic environment. The EKF updates the state variables with two types of measurement model. The first model is for the feature points and the second model is for the dynamic objects.