US 12,189,388 B2
Multiple inertial measurement unit sensor fusion using machine learning
Andrew Stewart, Chanhassen, MN (US); Christopher J. Mauer, Roseville, MN (US); Shashank Shivkumar, St. Louis Park, MN (US); and Thomas Jakel, Forrest Lake, MN (US)
Assigned to Honeywell International Inc., Charlotte, NC (US)
Filed by Honeywell International Inc., Charlotte, NC (US)
Filed on Jan. 5, 2022, as Appl. No. 17/569,316.
Prior Publication US 2023/0213936 A1, Jul. 6, 2023
Int. Cl. G06N 20/00 (2019.01); G01P 15/02 (2013.01); G05D 1/00 (2024.01)
CPC G05D 1/0088 (2013.01) [G01P 15/02 (2013.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A method, comprising:
receiving a plurality of inertial measurement data sets from a plurality of inertial sensors;
extracting features from the plurality of inertial measurement data sets;
training a first machine learning algorithm in multiple machine learning algorithms to select inertial measurement data sets in the plurality of inertial measurement data sets and apply weights to the selected inertial measurement data sets based on the extracted features;
training a second machine learning algorithm in the multiple machine learning algorithms to compensate and fuse the selected inertial measurement data sets; and
storing at least one fusion model produced by the multiple machine learning algorithms for application to a plurality of inertial measurements produced by the plurality of inertial sensors when tracking movement of a navigating object.