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].