NDPP-Grasp: Non-Differentiable Physical Plausibility Constraint-Guided Task-Oriented Dexterous Grasp Generation

2026-06-01

AI Summary

This work addresses a key limitation in existing task-oriented dexterous grasping methods based on diffusion models, which typically decouple task alignment from physical plausibility, often yielding kinematically or dynamically infeasible grasp trajectories. To overcome this, the authors propose a novel framework that directly embeds non-differentiable physical feasibility constraints into the denoising process of the diffusion model, enabling joint guidance by both task objectives and physical realism. This approach represents the first effective integration of non-differentiable physical constraints within diffusion-based trajectory generation, circumventing the drawbacks of conventional post-processing strategies. Experimental results demonstrate that the proposed method significantly improves both the physical plausibility and task suitability of generated grasps, outperforming current decoupled approaches.

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.

Problem

• task-oriented grasp
• dexterous grasp generation
• physical plausibility
• diffusion model
• non-differentiable constraints

Innovation

• non-differentiable constraints
• physical plausibility
• diffusion model
• task-oriented grasping
• dexterous grasp generation