Target Tracking for Robot-Fishes Based on Unscented Kalman Filter

Abstract:

Because of the nonlinear motion of the robot-fish and the uncertain water waves, it is difficult to track dynamic target accurately and quickly for the robot-fish. To solve this problem, this paper proposed the target tracking strategy of underwater robot-fish based on Unscented Kalman filter (UKF). UKF is a nonlinear filtering on the basis of unscented transformation, which consists of prediction and update recursively, thus estimate the state of the system. To avoid the particle degeneracy, this paper selects sigma points adopting scale symmetric sampling strategy. We apply UKF method to the underwater robot-fish competition in China. Simulations show that the Robot-fish has fast tracking speed, shorter tracking path and smaller tracking error using UKF method than not using the method. It has excellent tracking performance.

• Periodical: Applied Mechanics and Materials (Volumes 536-537)
• Pages: 979-983
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.536-537.979
• Online since: April 2014
• Authors: Peng Fei Bai, Yu Lian Jiang*, Rong Fang Lei
• Keywords: Target Tracking, Underwater Robot-Fishes, Unscented Kalman Filter

References

[1] Weisheng Yan , Lei Gao, Bo Zhang and Peixian Duan: The technology of underwater target tracking (2000 )in chinese.

[2] Hongren Zhou , Zhongliang Jing and Peide Wang: Maneuvering Target Tracking (1991)in chinese.

DOI: 10.1109/cdc.1990.204028

[3] Intelligent Control Laboratory of Peking University China Underwater robot 2D simulation group competition rules (2012).

[4] S.J. Julier and J.K. Uhlmann: Unscented filtering and nonlinear estimation (2004).

[5] S. J. Julier: A new method for the nonlinear transformation of means and covariance in filters and estimators(2000).

[6] S. J. Julier: The scaled unscented transformation(2002).

[7] R. E. KALMAN: A New Approach to Linear Filtering and Prediction Problems (1960).