English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

6D pose recognition has been a crucial factor in the success of roboticgrasping, and recent deep learning based approaches have achieved remarkableresults on benchmarks. However, their generalization capabilities in real-worldapplications remain unclear. To overcome this gap, we introduce 6IMPOSE, anovel framework for sim-to-real data generation and 6D pose estimation. 6IMPOSEconsists of four modules: First, a data generation pipeline that employs the 3Dsoftware suite Blender to create synthetic RGBD image datasets with 6D poseannotations. Second, an annotated RGBD dataset of five household objectsgenerated using the proposed pipeline. Third, a real-time two-stage 6D poseestimation approach that integrates the object detector YOLO-V4 and astreamlined, real-time version of the 6D pose estimation algorithm PVN3Doptimized for time-sensitive robotics applications. Fourth, a codebase designedto facilitate the integration of the vision system into a robotic graspingexperiment. Our approach demonstrates the efficient generation of large amountsof photo-realistic RGBD images and the successful transfer of the trainedinference model to robotic grasping experiments, achieving an overall successrate of 87% in grasping five different household objects from clutteredbackgrounds under varying lighting conditions. This is made possible by thefine-tuning of data generation and domain randomization techniques, and theoptimization of the inference pipeline, overcoming the generalization andperformance shortcomings of the original PVN3D algorithm. Finally, we make thecode, synthetic dataset, and all the pretrained models available on Github.

6IMPOSE: Bridging the Reality Gap in 6D Pose Estimation for Robotic Grasping