DAGDiff: Guiding Dual-Arm Grasp Diffusion to Stable and Collision-Free Grasps

Md Faizal Karim, Vignesh Vembar, Keshab Patra, Gaurav Singh, K Madhava Krishna · 2025

Read DAGDiff: Guiding Dual-Arm Grasp Diffusion to Stable and Collision-Free Grasps by Md Faizal Karim, Vignesh Vembar, Keshab Patra, Gaurav Singh, K Madhava Krishna (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-21145.txt.

Description

Two robot arms grasping one large object must agree on holds that are stable and do not collide. Md Faizal Karim and four co-authors present DAGDiff in a 2025 paper on dual-arm grasping. Earlier methods proposed each arm's grasp separately, leaning on region hints or heuristics.

DAGDiff generates the grasp pair jointly, denoising directly in the combined space of both gripper poses. Classifier signals guide the diffusion toward grasps that satisfy geometry, stability, and collision constraints. Enforcing those properties during generation, rather than filtering afterward, is the key insight. The framework needs no explicit region detection or object priors. Reliable two-arm grasps matter for warehouses and factories handling objects too big or awkward for one arm.

Catalog record

CREATORMd Faizal Karim, Vignesh Vembar, Keshab Patra, Gaurav Singh, K Madhava Krishna
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-21145

Browse the full library · About the Tower of Babel