aiqbench analysis
RAGrasp introduces a retrieval-augmented pipeline that enables robots to adapt grasping strategies using locally collected templates without requiring full model retraining. By leveraging self-supervised visual features and Segment Anything Model 2, the system achieves high success rates on both familiar and novel objects in real-world trials. This approach addresses a critical bottleneck in embodied AI by facilitating rapid deployment-specific adaptation with minimal annotation overhead.
RAGrasp: Geometry-Semantic Template Retrieval and Grasp Transfer
arXiv:2610.04438v1 Announce Type: new Abstract: We present RAGrasp, a retrieval-augmented pipeline for planar parallel-jaw grasping from a compact set of locally collected, grasp-annotated RGB-D (color and depth) templates. Unlike task-specific predictors trained primarily on large public or synthetic grasp datasets, RAGrasp requires no end-to-end retraining for a new deployment.Its template memory is constructed from observations collected with the deployment camera, robot, and gripper in the target workspace, thereby aligning stored examples with the local sensing and embodiment conditions. The system uses self-supervised DINOv2 visual fea- tures together with appearance and depth cues to prompt the Segment Anything Model 2 (SAM2), which isolates the query object. A two-stage geometry-semantic retrieval cascade then selects a template, and a confidence gate chooses one of two grasp- transfer estimators. The transferred grasp is refined using mask- support and silhouette-contact constraints before calibrated 2D- to-3D conversion. In real-world trials, RAGrasp achieves 20/20 successful grasps on seen objects and 19/20 on unseen objects. Within the evaluated setting, the results demonstrate deployment- specific grasp adaptation from limited local annotation and tolerance to the tested viewpoint and illumination changes.
Abstract:We present RAGrasp, a retrieval-augmented pipeline for planar parallel-jaw grasping from a compact set of locally collected, grasp-annotated RGB-D (color and depth) templates. Unlike task-specific predictors trained primarily on large public or synthetic grasp datasets, RAGrasp requires no end-to-end retraining for a new this http URL template memory is constructed from observations collected with the deployment camera, robot, and gripper in the target workspace, thereby aligning stored examples with the local sensing and embodiment conditions. The system uses self-supervised DINOv2 visual fea- tures together with appearance and depth cues to prompt the Segment Anything Model 2 (SAM2), which isolates the query object. A two-stage geometry-semantic retrieval cascade then selects a template, and a confidence gate chooses one of two grasp- transfer estimators. The transferred grasp is refined using mask- support and silhouette-contact constraints before calibrated 2D- to-3D conversion. In real-world trials, RAGrasp achieves 20/20 successful grasps on seen objects and 19/20 on unseen objects. Within the evaluated setting, the results demonstrate deployment- specific grasp adaptation from limited local annotation and tolerance to the tested viewpoint and illumination changes.
Artificial Intelligence (cs.AI); Robotics (cs.RO)
arXiv:2610.04438 [cs.AI]
(or arXiv:2610.04438v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.04438 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 274 words
Related coverage
J++ Lens: Jacobian Filtering Enables More Faithful Workspace Lenses
LessWrong · Oct 8, 2026
DualManip: Agentic Dynamic Manipulation via Dual-Path Semantic Reasoning and Geometric Adaptation
arXiv · Sep 28, 2026
A Storage-Retrieval Gap in Parametric Knowledge Graph Memory
arXiv · Aug 27, 2026
InsightEmb: Learning Action-Intent Embeddings for Agentic Insight Retrieval
arXiv · Aug 6, 2026