US Patent 11763525 Blind object tracking using point clouds

In one implementation, a method of blind object tracking is performed at a device including one or more processors and non-transitory memory. The method includes obtaining a first three-dimensional scene model of a physical environment during a first time period including a first plurality of points, wherein each of the first plurality of points is associated with a set of coordinates in a three-dimensional space, wherein a subset of the first plurality of points is associated with a particular cluster identifier. The method includes obtaining a second three-dimensional scene model of the physical environment during a second time period subsequent to the first time period including a second plurality of points, wherein each of the second plurality of points is associated with a set of coordinates the three-dimensional space. The method includes associating a subset of the second plurality of points with the particular cluster identifier, wherein the subset of the second plurality of points includes points with different sets of coordinates in the three-dimensional space than the points of the subset of the first plurality of points.