This dataset is collected from a portal crane at a production facility and is used to train a Convolutional Neural Network for accurate segmentation of grapples and headblocks. This dataset comprises 175 point cloud data derived from an Ouster LiDAR sensor (with 128 x 2048 resolution) mounted on the cabin of the crane. It contains several short time sequences, with each point cloud extracted and can be treated as a completely independent scene. To enhance the diversity of our dataset, we captured data featuring various grapple angles and heights, reflecting its dynamic movement during production. We additionally included scenarios where the grapple was occluded while grasping or depositing wood bundles.
To encourage further research and development in the field of crane automation, we have made this dataset publicly available. As there is a significant lack of publicly available datasets specifically tailored for portal cranes and their unique grapple shapes, we hope this dataset will be a valuable resource for the research community.
Portal Crane Dataset © 2025 by PSIORI GmbH is licensed under CC BY-SA 4.0