In China, our client was attempting to deploy an autonomous solution for an automobile flywheel production line. This entails the recognition of the location and orientation of the flywheels before they can be picked and then placed.
Due to a wide range of object poses and difficult object reflectiveness and tilting angle, this task is quite difficult for a pick-and-place system. In addition, there was a significant space constraint for the robot to avoid the box which holds the flywheels.
Utilizing our software system in conjunction with the BP-L camera, the extremely reflective flywheels could be detected using highly-accurate point cloud models, resulting in 3D box picking application scenarios with high precision, speed, and a large field of vision.
Our system was able to determine the orientation of each flywheel based on the detection of a feature only 2mm in size, which was necessary for the placing operation.
The solution has now been deployed on their workstation for a year with perfect uptime.