Abstract
The estimation of a cameras egomotion during image acquisition is a mandatory task for many different computer vision applications such as Structure from Motion (SfM), Simultaneous Localisation and Mapping (SLAM) or Augmented Reality (AR). The vast majority of the proposed applications are deriving the motion parameters indirectly from the captured images. This paper suggests a smart sensor system (S3) composed from three different micro-electromechanical (MEMS) inertial sensor types as an aiding modality for visionbased camera pose estimation. The S3 implementation contains a signal conditioning unit and a bank of Kalman filters for orientation estimation. The whole system is evaluated by using an industrial robot for the generation of specific motion patterns and the corresponding ground truth orientation measurements.