Open AccessBinocular Vision Pose Estimation Based on PSOPF
Author(s) -
F. Richard Yu,
Qiang Fu,
Yanyan Zhuang,
Xiang Liu,
Gonglei Liao,
Qichun Zhao
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1302/3/032051
Subject(s) - computer vision , artificial intelligence , pose , scale invariant feature transform , particle filter , binocular vision , computer science , robustness (evolution) , robot , 3d pose estimation , monte carlo localization , monocular vision , filter (signal processing) , feature extraction , biochemistry , chemistry , gene
In order to realize robot autonomous localization and navigation, binocular vision localization based on particle swarm optimized particle filter (PSOPF) is put forward according to the characteristics of nonlinear and non-Gaussian distribution complex system. The localization of 6-DOF robot only depends on binocular vision in the method. At first, road signs are obtained by SIFT feature matching points of binocular vision. The second, initial pose estimation is obtained by four elements. Finally, robot pose is estimated accurately by PSOPF, and the algorithm overco mes the shortcoming of particle filter (PF) and imp rove estimation accuracy. Experiment results show that this algorithm has high computing accuracy and robustness.