● 本报记者 郑萃颖 随着AI技术的影响力不断增强，空间智能技术成为解锁千行百业智能化升级不可或缺的能力。群核科 技将累积的物理正确的三维数据与AI技术结合，形成空间智能核心能力，并陆续应用于空间设计、工 业数字孪生、机器人仿真训练、内容创作等领域。 近期，群核科技持续拓展技术应用边界，构建起覆盖多场景的产品矩阵。通过开放底层能力与生态合 作，群核科技正以"空间智能卖水人"的定位，赋能AI技术在三维场景的规模化落地。 构建产品矩阵 群核科技近期发布新产品的节奏加快。据悉，本周群核科技孵化的3D AI内容创作工具LuxReal将启动小 范围内测。该创作工具主要面向电商从业者、影视剧编导、广告导演等群体的专业创作需求，能在保持 时空一致性的前提下，高效生成视频内容，避免了AI生成视频中，因切换视角带来的空间逻辑混乱等 常见问题。 在群核科技12月举办的2025酷+大会上，公司创始团队集中介绍了公司产品的最新进展，除了 LuxReal，还有此前推出的空间智能训练平台SpatialVerse、工业AI孪生平台SpatialTwin，以及酷家乐上 线的AI智能设计平台。 据介绍，SpatialVerse为具身智能训练提 ...

Core Viewpoint - The article discusses the launch of the InteriorGS dataset by Qunhe Technology, aimed at enhancing spatial perception capabilities for robots and AI agents, marking a significant advancement in AI training methodologies [2][4]. Group 1: InteriorGS Dataset - The InteriorGS dataset includes 1,000 3D Gaussian semantic scenes covering over 80 types of indoor environments, providing AI agents with a "spatial brain" to improve their environmental understanding and interaction capabilities [2][4]. - This dataset is claimed to be the world's first large-scale 3D dataset suitable for the free movement of intelligent agents [2][4]. Group 2: 3D Gaussian Technology - In recent years, 3D Gaussian technology has gained traction in fields such as cultural heritage preservation and spatial design due to its ability to "scan and reconstruct scenes" [3]. - Qunhe Technology engineers have utilized this technology to recreate a 60-year-old photo studio in Hangzhou, which has garnered significant attention [3]. Group 3: Training Data Generation Pathway - The dataset is part of a new training data generation pathway that combines 3D Gaussian reconstruction, spatial large model capabilities, and physical simulation [6]. - This process allows for rapid scene reconstruction from videos or images, enriching the data with semantic logic and simulating physical characteristics for intelligent agents to learn spatial understanding and interaction [6]. Group 4: Spatial Intelligence Platform - Qunhe Technology's SpatialVerse is positioned as a leading spatial intelligence training platform, having accumulated vast amounts of interactive 3D data and a suite of physical simulation tools [9]. - The goal of SpatialVerse is to become the "ImageNet" of the spatial intelligence field, similar to how ImageNet catalyzed the explosion of computer vision [9]. Group 5: Industry Collaboration - Qunhe Technology has formed partnerships with several embodied intelligence companies, including Zhiyuan Robotics and Galaxy General, to leverage its high-quality 3D scene data [9]. - The company aims to advance the "Sim2Real" paradigm, which is currently the most efficient training method for embodied intelligence, by collaborating with industry players [9].