Three dimensional low-speed motion tracking using micro inertial measurement unit and monocular visual sensor

We present in this paper a fusion method of combining vision and inertial (accelerations and angular velocities) data for estimating and predicting position and orientation (pose) of a rapidly moving camera with respect to a fixed inertial frame. The basic framework of this fusion method is based on the Kalman filtering algorithm. By fusing the data, a fast, accurate and robust pose estimation is obtained. Moreover, the fusion system can provide a reference for micro inertial measurement unit (μIMU) in order to eliminate drift errors due to μIMU´s intrinsic biases and random noise such as circuit thermal noise. In order to evaluate the performance of the system, an experiment was conducted and the results are summarized and discussed in this paper.