Workflow
具身智能虚拟训练平台
icon
Search documents
索辰科技:物理AI为具身智能系统提供更高的稳定性、可解释性和现实适应性
证券日报网12月17日讯 索辰科技在12月16日回答调研者提问时表示,物理AI通过将物理规律融入AI模 型中,为具身智能系统提供更高的稳定性、可解释性和现实适应性。2025年7月索辰发布具身智能虚拟 训练平台,平台具备高效的算法迭代加速能力。通过物理AI核心体系与计算物理引擎的深度融合,平 台能精准复刻物理世界的各项参数,从物体的材质、重力的影响到光线的变化,从外部气流变化到电磁 干扰感知,从结构接触响应到电池能源管理,都与真实环境高度一致;将真实世界感知数据与虚拟世界 训练数据链路无缝衔接,实现真正意义的虚实交融,避免因虚拟与现实差异过大而导致的训练成果失效 问题,彻底打通真实世界与虚拟世界的数据壁垒,实现"采集-训练-生成-评估"的完整链路,让机器人训 练更经济、更高效、更真实、更简单。 (编辑 袁冠琳) ...
索辰科技董事长陈灏:物理AI有望成为新一代AI核心范式
Core Insights - The core focus of the articles is on the advancements in physical AI technology by Suochen Technology, which aims to revolutionize the AI landscape by integrating physical laws and real-world data into AI decision-making processes [1][2][4]. Group 1: Physical AI Development - Suochen Technology has launched four virtual training platforms powered by physical AI, applicable in humanoid robot training, low-altitude economy 3D physical mapping, and virtual training for new energy batteries [1]. - The company has invested in physical simulation laboratories in Jiaxing, Zhejiang, covering an area of 40,000 square meters to validate and implement physical AI technology [1]. - Physical AI is seen as a new paradigm in AI, allowing AI to deduce and make decisions like human scientists, potentially leading to disruptive value even with minimal breakthroughs [1][2]. Group 2: Cost Efficiency and Technological Integration - Physical AI offers significant cost advantages; for instance, traditional CPU calculations for vehicle simulations may take days, while using consumer-grade GPUs can reduce this to a few hours, potentially lowering costs to 1% or even 0.1% of traditional methods [3]. - The technology combines computational physics, mathematics, computer graphics, and GPU capabilities, positioning it as a core technology for transforming industrial equipment [2][3]. Group 3: Strategic Partnerships and Market Position - Suochen Technology has signed a strategic cooperation agreement with Hangzhou Industrial Investment Group to provide high-precision simulation and virtual training support for low-altitude infrastructure projects [6]. - The company is actively pursuing acquisitions to fill gaps in data collection and engineering, including a recent acquisition of 55% of Ningbo Maishijie Technology for 88 million yuan [7]. - The establishment of a Hungarian subsidiary will serve as a strategic platform for overseas acquisitions, indicating a focus on expanding its market presence [8]. Group 4: Industry Impact and Future Outlook - The integration of physical AI is expected to lead to a paradigm shift in various industries, with the potential for significant upgrades in CAE (Computer-Aided Engineering) technologies [8]. - Suochen Technology aims to continue exploring the application of physical AI to empower industrial equipment design and manufacturing, facilitating rapid development for clients [8].