Method of performing simultaneous localization and mapping with respect to a salient object in an image

A cloud server,

• configured to receive, from a robot, one or more images for the robot to perform simultaneous localization and mapping (SLAM) with respect to a salient object for estimating a location of the robot within a space, comprising:
• a non-transitory map storage configured to store information for the robot to perform the SLAM;
• and
• a server controller configured to:
• detect an object from the received one or more images;
• select, as a specific salient object for identifying the space, a detected object verified as corresponding to the specific salient object;
• store, in the map storage, the selected specific salient object and coordinate information related to the selected specific salient object, wherein the coordinate information includes information on a position and a size of a bounding box around the selected specific salient object;
• and
• predict a next position of the selected specific salient object using multiple-object tracking in subsequent images when the selected specific salient object is extracted from a first image among the received one or more images.

A method comprising:

• causing a robot to move within a space;
• capturing, by a camera sensor of the robot, one or more images for performing a simultaneous localization and mapping (SLAM) with respect to a salient object for estimating a location of the robot within the space;
• detecting an object from the one or more captured images;
• selecting, as a specific salient object for identifying the space, a detected object verified as corresponding to the specific salient object;
• storing the selected specific salient object and coordinate information related to the selected specific salient object, wherein the coordinate information includes information on a position and a size of a bounding box around the selected specific salient object;
• and
• predicting a next position of the selected specific salient object using multiple-object tracking in subsequent images when the selected specific salient object is extracted from a first image among the one or more captured images.