We consume 3D Point Cloud data and turn it into a Software solutions.
Our software approach is predicated on the collection and use of real world data throughout the development cycle. This approach reveals all of the challenges faced by vehicles operating in the "messy" material handling industry. For PDS, this means working with pallets that are damaged, pockets that are covered with shrink wrap, and all other manner of imperfections that can potentially lead to a false positive or negative pallet detection.
PDS is tested against a curated set of 400 real world pallet examples. This set is designed to verify that the PDS solution will operate correctly in all environments and against even the most challenging pallet conditions.
Generate a full 3 dimensional point cloud
The PDS detection starts with capturing the full amplitude and distance data from the O3D. These images can contain noise and artifacts due to the harsh environmental conditions.
Filter the image
The full image is filtered to “clean" the image and eliminate unwanted pixels. This is a critical step in bringing robust pallet detection.
Determine 6-DoF pose
The filtered image is then used to determine the 6-DoF pose of the pallet and its pockets.
This approach was taken in order to robustly achieve the following PDS specifications: