An Application of Unscented Kalman Filter for Pose and Motion Estimation Based on Monocular Vision

In a pose and motion estimation system from monocular vision, the extended Kalman filter (EKF) is a widely used filtering strategy generally. However, as defects of the EKF in nonlinear estimation, there exists estimated error, which affects the accuracy of the state estimation, when it is adopted in nonlinear estimation. In order to yield the higher accuracy of pose and motion estimation, the unscented Kalman filter (UKF) using the principle that a set of discretely sampled points can be used to parameterize mean and covariance was employed in this paper. Given a sequence of 2D monocular images of an moving object, using line features on the image plane as measured inputs and a dual quaternions to represent the 3D transformation, the indirect measurement solutions of pose and motion estimation from monocular vision is presented based on EKF and UKF with simulated data. Simulation results have shown that the UKF is a superior alternative to the EKF

[1]  Kostas Daniilidis,et al.  Hand-Eye Calibration Using Dual Quaternions , 1999, Int. J. Robotics Res..

[2]  Jeffrey B. Burl,et al.  A reduced order extended Kalman filter for sequential images containing a moving object , 1993, IEEE Trans. Image Process..

[3]  Gaurav S. Sukhatme,et al.  Circumventing dynamic modeling: evaluation of the error-state Kalman filter applied to mobile robot localization , 1999, Proceedings 1999 IEEE International Conference on Robotics and Automation (Cat. No.99CH36288C).

[4]  Stergios I. Roumeliotis,et al.  Augmenting inertial navigation with image-based motion estimation , 2002, Proceedings 2002 IEEE International Conference on Robotics and Automation (Cat. No.02CH37292).

[5]  Avinash C. Kak,et al.  Real-time tracking and pose estimation for industrial objects using geometric features , 2003, 2003 IEEE International Conference on Robotics and Automation (Cat. No.03CH37422).

[6]  Gregory D. Hager,et al.  Fast and Globally Convergent Pose Estimation from Video Images , 2000, IEEE Trans. Pattern Anal. Mach. Intell..

[7]  Jeffrey K. Uhlmann,et al.  New extension of the Kalman filter to nonlinear systems , 1997, Defense, Security, and Sensing.

[8]  David Nistér,et al.  An efficient solution to the five-point relative pose problem , 2003, 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003. Proceedings..

[9]  J.C.K. Chou,et al.  Quaternion kinematic and dynamic differential equations , 1992, IEEE Trans. Robotics Autom..

[10]  Mongi A. Abidi,et al.  A New Efficient and Direct Solution for Pose Estimation Using Quadrangular Targets: Algorithm and Evaluation , 1995, IEEE Trans. Pattern Anal. Mach. Intell..

[11]  R. Chellappa,et al.  Recursive 3-D motion estimation from a monocular image sequence , 1990 .

[12]  Jin Cao,et al.  Computer vision system for an autonomous mobile robot , 2000, SPIE Optics East.