Method and apparatus for simultaneous localization and mapping, and computer-readable storage medium

An apparatus for visual-inertial simultaneous localization and mapping (SLAM), the apparatus comprising:

• at least one processor configured to:
• detect features of points and lines from a current image frame;
• obtain a Manhattan frame of the current image frame using the detected features of the lines;
• construct a spatiotemporally-sensitive global sub map using a global map generated by visual-inertial SLAM;
• re-identify the Manhattan frame by re-identifying the points and lines using the spatiotemporally-sensitive global sub map;
• obtain a re-identification constraint with regard to a relationship among the points, the lines, and the Manhattan frame based on the re-identification;
• and
• obtain a SLAM result based on the re-identification constraint.

An electronic device comprising:

• a memory configured to store a computer program;
• and
• a processor configured to execute the computer program to implement a method comprising:
• detecting features of points and lines from a current image frame;
• obtaining a Manhattan frame of the current image frame using the detected features of the lines;
• constructing a spatiotemporally-sensitive global sub map using a global map generated by a visual-inertial simultaneous localization and mapping (SLAM);
• re-identifying the Manhattan frame by re-identifying the points and lines using the spatiotemporally-sensitive global sub map;
• obtaining a re-identification constraint with regard to a relationship among the points, the lines, and the Manhattan frame based on the re-identification;
• and
• obtaining a SLAM result based on the re-identification constraint.

A non-transitory computer-readable storage medium having stored thereon a computer program which, when executed, implements a method comprising:

• detecting features of points and lines from a current image frame;
• obtaining a Manhattan frame of the current image frame using the detected features of the lines;
• constructing a spatiotemporally-sensitive global sub map using a global map generated by a visual-inertial simultaneous localization and mapping (SLAM);
• re-identifying the Manhattan frame by re-identifying the points and lines using the spatiotemporally-sensitive global sub map;
• obtaining a re-identification constraint with regard to a relationship among the points, the lines, and the Manhattan frame based on the re-identification;
• and
• obtaining a SLAM result based on the re-identification constraint.