Ran Ding

MessyKitchens: Contact-rich object-level 3D scene reconstruction

Junaid Ahmed Ansari*, Ran Ding*, Fabio Pizzati, Ivan Laptev

* Co-first authorsarXiv, 2026

Project Page arXiv

Monocular 3D scene reconstruction has recently seen significant progress, but reconstructing cluttered scenes into physically plausible 3D objects with accurate contacts remains challenging. We introduce MessyKitchens and a Multi-Object Decoder to improve object-level scene reconstruction with better registration accuracy and inter-object penetration.

3D-Aware Diffusion Policy for Coverage Path Planning

Chenyuan Chen, Haoran Ding, Ran Ding, Tianyu Liu, Zewen He, Anqing Duan, Yoshihiko Nakamura

arXiv, 2026

3D-CovDiffusion introduces a 3D-aware diffusion policy for coverage path planning, aiming to improve coverage behavior through stronger 3D geometric understanding.

TARGO: Benchmarking Target-driven Object Grasping under Occlusions

Ran Ding*, Ziyuan Qin*, Yan Xia*^†, Guanqi Zhan, Kaichen Zhou, Long Yang, Hao Dong, Daniel Cremers

* Co-first authors, ^† Corresponding authorInternational Journal of Computer Vision, 2026

Project Page arXiv

This work introduces TARGO, a benchmark dataset for target-driven grasping under occlusions in cluttered environments, along with TARGO-Net to improve grasping performance in these physically challenging settings.