This paper presents and compares the performance of nonlinear estimation filters for the inertial SLAM (Simultaneous Localization and Mapping) integrated navigation system including the extended Kalman filter, the unscented Kalman filter and the particle filter. A computer simulation is conducted to analyze the navigation accuracy as well as the capability of real-time implementation by individual filter using a monocular vision based navigation model. The detail model for the linear filter design and the initial delayed localization of the target features were investigated. Simulation results show that the unscented Kalman filter has better performance in perspective of both the navigation performance and the feasibility of real-time implementation.
Digital Object Identifier (DOI)
Hee Won, Dae; Yun, Sukchang; Jae Lee, Young; and Sung, Sangkyung
"System Modeling and Non-Linear Estimation Performance Comparison of Monocular Vision based Integrated Navigation System,"
Applied Mathematics & Information Sciences: Vol. 10:
2, Article 5.
Available at: https://digitalcommons.aaru.edu.jo/amis/vol10/iss2/5