US 12,327,321 B2
Automatic detection of the presence of a moving platform
David John McKinnon, Redmond, WA (US); Joshua Aidan Elsdon, Bellevue, WA (US); Max S Kaznady, Arlington, MA (US); and Evan Gregory Levine, Seattle, WA (US)
Assigned to Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed by Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed on Dec. 29, 2022, as Appl. No. 18/090,712.
Prior Publication US 2024/0221322 A1, Jul. 4, 2024
Int. Cl. G06T 7/20 (2017.01); G06T 19/00 (2011.01)
CPC G06T 19/006 (2013.01) [G06T 7/20 (2013.01); G06T 2207/20081 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method for using a trained predictive machine learning (ML) algorithm to use convoluted motion data to inferentially determine a category for a moving platform on which a mixed-reality (MR) system is operating, the convoluted motion data comprising at least a first signal and a second signal, and the trained predictive ML algorithm determines the category without decomposing the convoluted signal, said method comprising:
detecting a display artifact that is associated with content displayed by the MR system;
determining that a current configuration of a motion model used to display the content is causing the display artifact;
analyzing a time-limited series of convoluted motion data, wherein the time-limited series of convoluted motion data includes first motion data representing a motion of the MR system and second motion data representing a motion of the moving platform, and wherein the first motion data is convoluted with the second motion data to form the time-limited series of convoluted motion data;
accessing the trained predictive ML algorithm, which is trained to categorize moving platforms using convoluted motion data without decomposing the convoluted motion data into its constituent motion data components;
feeding the time-limited series of convoluted motion data as input to the predictive ML algorithm;
causing the predictive ML algorithm to determine a particular category for the moving platform based on the time-limited series of convoluted motion data; and
based on the determined category, triggering, in real time, either (i) use of a reconfigured version of the motion model or (ii) use of a new motion model.