Position sensing using integration of a vision system and inertial sensors

Abstract

This paper concerns the development of a position tracking system for a hand-held tool based on the low cost sensors. The integration of a multi-camera vision system and a strapdown inertial navigation system using indirect Kalman filter (IKF) is able to compute the 3D position of a tool tip, which is the point of interest for tracking, without prior knowledge of the motion. The continuous linear and angular motion of the tool is sensed in 3D by using MEMS-based inertial sensors. At the same time, the tool is tracked by a multi-camera vision system. The multi- camera vision system includes four low cost CCD cameras, when all four cameras are configured to be placed on a curved line instead of the classical arrangement. The experimental results show that the position errors of the tool tip tracking based on the proposed vision system are decreased. Furthermore, the inertial sensors integrated with the vision system allow tracking an object with lower sampling rate than the vision system alone without loosing the accuracy.