PulseAugur
EN
LIVE 12:28:46
中文(ZH) ICRA 2026|港中文GeoLanG :基于几何感知的语言引导抓取技术,结合统一的 RGB-D 多模态学习机制

GeoLanG framework enhances robot grasping with geometric and language integration

Researchers have developed GeoLanG, a novel framework that unifies geometric reasoning with language understanding for robot grasping. This approach integrates RGB-D data and language features into a shared representation, with a depth-guided geometric module that incorporates geometric priors directly into the attention mechanism. GeoLanG has demonstrated superior performance on benchmarks and robustness in cluttered, occluded environments, with successful validation on real robot hardware. AI

IMPACT This framework could significantly improve robot manipulation capabilities in complex, real-world scenarios.

RANK_REASON The cluster describes a new research framework and its performance on benchmarks, fitting the research bucket. [lever_c_demoted from research: ic=1 ai=1.0]

Read on 雷峰网 (Leiphone) →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

GeoLanG framework enhances robot grasping with geometric and language integration

COVERAGE [1]

  1. 雷峰网 (Leiphone) TIER_1 中文(ZH) ·

    ICRA 2026 | CUHK GeoLanG: Geometric Perception-based Language-Guided Grasping Technology, Combining Unified RGB-D Multimodal Learning Mechanism

    <p style="margin-top: 0; margin-bottom: 0;"></p><p style="margin-top: 0; margin-bottom: 0;">原文链接:http://www.labren.org/mm/news/?-icra-2026-???????-???????/</p><p>&nbsp;</p><p style="margin-top: 0px; margin-bottom: 0px; text-align: center;"><img border="0" src="https://static.leip…