AlphaSpace：通过语义标记化与符号推理实现机器人行为

AlphaSpace: Enabling Robotic Actions through Semantic Tokenization and Symbolic Reasoning

March 24, 2025

作者: Alan Dao, Dinh Bach Vu, Bui Quang Huy

cs.AI

摘要

本文介绍了一种名为AlphaSpace的创新方法，旨在增强大型语言模型（LLMs）在三维笛卡尔空间导航中的空间推理能力。AlphaSpace采用基于语义的分词策略，通过专门的语义标记编码高度信息，并主要整合符号化的合成推理数据。该方法使LLMs能够通过将物体定位在特定的[x, y, z]坐标上，精确地操控物体。实验结果表明，AlphaSpace在操作子任务上显著优于现有模型，总体准确率达到66.67%，而GPT-4o和Claude 3.5 Sonnet的准确率分别为37.5%和29.17%。

English

This paper presents AlphaSpace, a novel methodology designed to enhance the spatial reasoning capabilities of large language models (LLMs) for 3D Cartesian space navigation. AlphaSpace employs a semantics-based tokenization strategy, encoding height information through specialized semantic tokens, and integrates primarily symbolic synthetic reasoning data. This approach enables LLMs to accurately manipulate objects by positioning them at specific [x, y, z] coordinates. Experimental results demonstrate that AlphaSpace significantly outperforms existing models on manipulation subtasks, achieving a total accuracy of 66.67%, compared to 37.5% for GPT-4o and 29.17% for Claude 3.5 Sonnet.