A wearable robot control interface based on measurement of human body motion using a camera and inertial sensors

Abstract

This paper describes a wearable robot control interface based on measurement of human body motion, which allows a user to control the robot intuitively. This wearable interface is composed of a camera and inertial sensors and estimates body motion of the user: movement and arm motion. The estimated motion is then converted into commands of movement or arm motion of a humanoid robot. The body motion estimation is performed by a combination of a monocular SLAM and an EKF-based motion estimation using inertial sensors. The arm motion is estimated with a Wii Remote with a visual marker using the same motion model and sensor data integration. We implemented efficient algorithms, especially for image processing, for a real-time motion estimation. We conducted motion estimation and robot control experiments, which show the effectiveness of the proposed method.