The Part Detection feature in Gocator analyzes a 3D point cloud created from either laser profile or fringe projection scans to automatically identify and separate discrete objects. The discrete objects or parts can then undergo correction for rotation to establish a common orientation ready for slicing and feature measurement.

Sophisticated Built-in Detection Logic

Gocator’s unique Part Detection feature is able to auto detect the beginning and end of parts and track multiple parts appearing simultaneously.

Typically, part detection is performed outside of a sensor by using an SDK on a PC and adding blob processing software. With Gocator, these capabilities are built-in and tightly coupled with the acquisition engine––processing on-the-fly as scans are developed into point clouds.

Tracking Multiple Parts

Using Part Detection, multiple parts can pass through the laser at the same time and will be individually tracked. Parts can be separated along the laser line (X axis), in the direction of travel (Y axis), by height thresholding or external input.

Gap Length and Gap Width to Connect Related Parts

Gocator’s Gap Length and Gap Width settings allow users to combine similar parts into a single part as they pass under the sensor.

Gap Width determines the minimum separation between objects on the X axis. Gap Length determines the minimum separation between objects on the Y axis. If parts are closer than the gap interval, they are merged into a single part.

Height Thresholding

Height Threshold controls the profile height threshold for part detection. The setting for Threshold Direction controls whether parts should be detected above or below the threshold. Above is typically used to prevent the belt surface from being detected while the object moves along a conveyor (eg., tracking bread dough and not the crumbs).

Gated External Input

An external input can be used to control when the sensor should scan to build a part. For example, a photocell is typically used in cases where a part travels past a photo detector that triggers scanning to capture the part. Another scenario is when a robot is moving the sensor past a feature and activates the external input to capture the feature as a part scan.

Future Enhancements

Part Detection handles collecting scans to form discrete objects for downstream measurement which often involves part rotation, slicing, and profile processing. With Gocator, many inspection tasks are easily solved with this built-in capability. In the future, separation of touching parts (like loaves of bread or plastic beads) and a count function (how many loaves of bread or how many beads were found) is planned to handle other part segmentation tasks.