NDPP-Grasp: Non-Differentiable Physical Plausibility Constraint-Guided Task-Oriented Dexterous Grasp Generation
2026-06-01 • Robotics
Robotics
AI summaryⓘ
The authors address the problem of generating grasp poses for robotic hands that are both physically realistic and suitable for specific tasks. Current methods often generate task-aligned grasps first and then try to fix any physical problems afterward, which can lead to less-than-ideal results. To improve this, the authors introduce a new method that considers physical plausibility during the entire grasp generation process, even when these constraints are hard to model mathematically. Their experiments show that this integrated approach works better than previous ones.
dexterous graspingtask-oriented graspingdiffusion modelsphysical plausibilitydenoising processgrasp refinementrobotic manipulationnon-differentiable constraints
Authors
Qiuchi Xiang, Haoxuan Qu, Hossein Rahmani, Jun Liu
Abstract
Task-oriented dexterous grasp generation aims to produce dexterous grasp poses that are both physically plausible and functionally suitable for specified manipulation tasks. Existing diffusion-based methods often address these two requirements in a decoupled manner: they first train a grasp diffusion model for task alignment and then rely on post-generation refinement to improve physical plausibility. However, this after-the-fact correction strategy applies physical plausibility guidance only once the grasp has already been generated, leaving the generation trajectory itself unguided by physical constraints and potentially leading to suboptimal grasps. To address this problem, we propose a novel framework that directly injects physical plausibility guidance into the denoising process of a task-aligned grasp diffusion model in a practical and effective manner, even when physical plausibility constraints are non-differentiable. This allows physical plausibility to shape grasp generation throughout denoising while preserving task alignment. Extensive experiments demonstrate the efficacy of our framework.