US 12,145,281 B2
Visual positioning method and system based on Gaussian process, and storage medium
Jianqiang Li, Guangdong (CN); Jie Chen, Guangdong (CN); Qinjian Li, Guangdong (CN); Shuqing Hu, Guangdong (CN); and Zhongming Liang, Guangdong (CN)
Assigned to SHENZHEN UNIVERSITY, Shenzhen (CN)
Filed by SHENZHEN UNIVERSITY, Guangdong (CN)
Filed on Feb. 7, 2022, as Appl. No. 17/666,550.
Application 17/666,550 is a continuation of application No. PCT/CN2020/127187, filed on Nov. 6, 2020.
Claims priority of application No. 202010731969.1 (CN), filed on Jul. 27, 2020.
Prior Publication US 2022/0168900 A1, Jun. 2, 2022
Int. Cl. B25J 9/16 (2006.01); G06N 7/01 (2023.01); G06V 10/42 (2022.01)
CPC B25J 9/1697 (2013.01) [B25J 9/163 (2013.01); B25J 9/1664 (2013.01); G06N 7/01 (2023.01); G06V 10/42 (2022.01)] 9 Claims
OG exemplary drawing
 
1. A visual positioning method based on a Gaussian process, comprising:
collecting image information of a surrounding environment and moving trajectory points while traveling;
extracting global features and semantic features in the collected image information;
processing the extracted global features and semantic features and the moving trajectory points according to a preset processing rule to obtain a Gaussian process observation model;
reconstructing a Bayes filtering framework according to the Gaussian process observation model, endowing a current trajectory with an initial position point, and generating a next position point of the current trajectory through the reconstructed Bayes filtering framework, the next position point being used for providing a positioning guidance for navigation; and
controlling a vehicle to traverse the current trajectory;
wherein the manners of extracting global features and semantic features in the collected image information respectively are:
extracting dimensions of global features in the collected image information through a Steerable Pyramid algorithm; and
extracting a maximum probability value of different categories of things in each collected picture through CenterNet algorithm semantics.