US 11,748,942 B2
System and method for automatically generating trajectories for laser applications
Chuong Luong, East Victoria Park (AU); Martin Fankhauser, Bruck (AT); Christopher Whiteside, Victoria (AU); Marcus Clarke, Mount Martha (AU); and Walter Meyler, Lysterfield (AU)
Assigned to Siemens Mobility Pty Ltd, Bayswater (AU); and Automation Innovation Pty Ltd, Keysborough (AU)
Filed by Siemens Mobility Pty Ltd., Bayswater (AU); and Siemens Ltd., Melbourne (AU)
Filed on Aug. 13, 2020, as Appl. No. 16/992,493.
Prior Publication US 2022/0048080 A1, Feb. 17, 2022
Int. Cl. G06T 17/00 (2006.01); G05B 19/19 (2006.01)
CPC G06T 17/00 (2013.01) [G05B 19/19 (2013.01)] 14 Claims
OG exemplary drawing
 
1. A system for automatically generating trajectories of an object comprising:
a trajectory generation module comprising a visual control algorithm and at least one processor configured via computer executable instructions to
receive raw three dimensional (3-D) sensor data of an object,
create a 3-D model of the object based on the raw 3-D sensor data,
extract object features relating to a shape and/or surface from the 3-D model of the object, and
generate trajectories based on the object features of the 3-D model of the object,
wherein the trajectory generation module further comprises a network interface, the trajectory generation module being further configured to send key data including the object features of the 3-D model to a remote database via the network interface,
wherein the remote database is configured to collect key data of multiple objects transmitted by multiple trajectory generation modules at different locations or sites,
wherein a remote application utilizes the key data of the multiple objects stored in the remote database to optimize the visual control algorithm by applying an unsupervised machine learning technique, and
wherein an optimized version of the visual control algorithm is deployed periodically to the multiple trajectory generation modules at the different locations or sites.