US 12,436,063 B1
Bearing ball inspection system
Lance Edelman, Atlanta, GA (US); Mahdi Bagheri, Duluth, GA (US); and Robert Alan Edelman, Farmington, UT (US)
Assigned to Vallum Software, LLC, Peachtree Corners, GA (US)
Filed by Vallum Software, LLC, Peachtree Corners, GA (US)
Filed on May 28, 2025, as Appl. No. 19/220,651.
Claims priority of provisional application 63/661,918, filed on Jun. 20, 2024.
Int. Cl. G01M 13/04 (2019.01); G06T 7/00 (2017.01); H04N 23/56 (2023.01); H04N 23/73 (2023.01); H04N 23/90 (2023.01)
CPC G01M 13/04 (2013.01) [G06T 7/0004 (2013.01); H04N 23/56 (2023.01); H04N 23/73 (2023.01); H04N 23/90 (2023.01); G06T 2207/20081 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A bearing ball inspection system comprising:
a rail guide configured to deliver a bearing ball to an inspection point, the rail guide comprising:
an inlet gate configured to rotate in a sweeping motion to release one bearing ball at a time toward the inspection point;
a funnel inlet configured to receive a plurality of bearing balls and direct each of the bearing balls toward the inlet gate; and
a ball aligner operatively coupled to an actuator and configured to guide the released bearing ball onto a positioning wheel for inspection;
a light source configured to illuminate the bearing ball;
a plurality of cameras positioned around the inspection point and configured to capture a first set of images of a first surface of the bearing ball;
the positioning wheel located at the inspection point and configured to support and rotate the bearing ball during inspection to reposition the bearing ball to expose a second surface;
the plurality of cameras further configured to capture a second set of images after repositioning;
a control system comprising a microprocessor, a memory, and a communication interface;
a machine learning model configured to analyze the first set of images and the second set of images to detect a defect on a surface of the bearing ball, wherein the analysis is performed by the control system, or by a connected computing device or a remote data processing resource via the communication interface; and
a sorting mechanism configured to direct the bearing ball to a fail egress if the defect is detected by the machine learning model, or to a pass egress if no defect is detected by the machine learning model.