本文来源:https://www.xhby.net/content/s6afecc3ae4b0393f8aeaa3e8.html
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。
近日,2026江苏人工智能创新发展大会暨第二届人工智能OPC大会成功举办,大会以“AI·成就梦想”为主题,集中展示了江苏人工智能领域的最新成果,重磅发布了苏州市支持人工智能创新发展“政策包”及一批人工智能领域创新成果和要素。
会上,俞凯发表《构建可持续创新的“模芯云用”开源生态》主旨演讲,围绕AI从数字世界迈向物理世界的产业变革,深度解读硬件终端具身智能体发展趋势,分享具智社区、Open-Bench评测平台建设成果,赋能江苏制造业智能化转型升级。
俞凯在演讲中指出,AI上半场聚焦数字世界内容生成、数据分析,而下半场的竞争主战场已经转向物理世界,具身智能成为全球大国科技竞争关键赛道。我国“十五五”规划纲要明确将具身智能列为全新经济增长点。
长三角拥有雄厚制造业底盘,叠加江苏省系列“人工智能+”政策红利,江苏具备发展硬件终端具身智能体得天独厚的产业土壤。当前行业虽软件开源丰富,但模型与数据缺少通往硬件端侧落地的桥梁,制约智能体在实体终端量产应用。
同时,OPC模式正在完成从软件OPC到硬件OPC的跨越。软件OPC是“人+AI工具”快速完成软件验证,而硬件OPC则是“人+AI+设备供应链”,要完成传感器适配、端侧模型嵌入式开发、原型制造、真实场景测试交付。硬件OPC创新难度更高,亟须“模芯云用”一体化,才能形成“系统集成性”的智能终端应用。
针对这一行业卡点,思必驰牵头打造国内首个聚焦“硬件终端+智能体”的垂直领域开源社区具身智能体社区(以下简称“具智社区”),构建“模芯云用”一体化可持续创新开源生态,补齐端侧智能落地短板。具智社区瞄准可量产的端侧智能产品方案,坚持端云协同、软硬结合、产业导向、自主可控,面向国产芯片、国产操作系统做深度适配,打通从算法原型到硬件产品落地的关键鸿沟。
会上,俞凯重点介绍具智社区Open-Bench平台,该平台是开放、可复现的AI实验评测平台,平台聚焦软硬件结合的具身智能与开源模型评测场景,可解决协议驱动的公平评测、统一框架的论文复现、可组合的智能体编排、端云一体的在线仿真等难题,打通智能体落地的完整闭环,从云端评测到虚拟物理场景在线仿真,形成可持续迭代的反馈回路,帮助科研团队、中小企业降低软硬件适配、原型开发成本,大幅缩短智能终端产品研发周期,推动具身机器人、智能制造等方案从概念验证走向真实业务部署。
大会期间,江苏省副省长沈剑荣一行莅临思必驰展区,俞凯向沈剑荣一行汇报了思必驰具身智能体开源社区建设情况及智慧办公、智慧物联等领域的创新成果。
本次参展的思必驰智慧办公产品包括吸顶麦系列、AI办公本系列、AI摄像头、录音卡、AI超级笔等硬件产品。
其中,面向政企办公推出的吸顶麦系列搭配AI摄像头,可构建视听清晰的本地+远程办公体验,解决会议中拾音不清晰、混响严重、视频看不清板书等难题。面向个人用户推出的AI办公本系列、AI录音卡、AI超级笔则主要解决个人办公常见难题。AI办公本面向深度会议场景,支持AI洞察,会后可给出行动建议,实现从“记录”到“决策”的功能跃迁;AI录音卡机身轻薄极致便携,补足移动与碎片场景下的拾音入口;AI超级笔支持手写+录音,可通过点选纸面笔迹快速定位到相应语音,三款产品覆盖会议座谈、移动采集、手写记录等常见个人办公场景。
在智慧物联领域,思必驰展出了具身机器人解决方案——天机Fusion多模态大模型交互系统。该系统使机器人不再局限于“听懂说话”,而是能够看懂环境、识别肢体与情绪,思考并执行相应动作,真正实现从“能听会说”到“会协同、懂情绪”的跨越,可适配人形机器人、服务机器人、扫地机器人、零售机器人、展厅讲解机器人等多种形态。
未来,思必驰将依托具智社区开源平台,推动AI技术深度赋能实体经济,助力新一代智能终端与智能体规模化普及,为江苏乃至全国人工智能产业高质量发展持续输出开源创新力量。
本文来源:https://www.xhby.net/content/s6afecc3ae4b0393f8aeaa3e8.html
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。
Recently, the 2026 Jiangsu Artificial Intelligence Innovation and Development Conference and the Second Artificial Intelligence OPC Conference were successfully held. The conference, themed "AI – Achieving Dreams," showcased the latest achievements in Jiangsu's artificial intelligence sector. It also featured the significant release of Suzhou's "policy package" supporting artificial intelligence innovation and development, along with a series of innovative achievements and key elements in the field of artificial intelligence.
At the conference, Yu Kai delivered a keynote speech titled "Building a Sustainable and Innovative 'Model-Chip-Cloud-Application' Open-Source Ecosystem." He focused on the industrial transformation of AI from the digital world to the physical world, provided an in-depth analysis of the development trends of embodied intelligent agents in hardware terminals, and shared the achievements of the Embodied Intelligence Community and the Open-Bench evaluation platform, thereby empowering the intelligent transformation and upgrading of Jiangsu's manufacturing industry.
In his speech, Yu Kai pointed out that the first half of AI focused on content generation and data analysis in the digital world, while the main battleground for competition in the second half has shifted to the physical world, with embodied intelligence becoming a key arena in the technological competition among global powers. China's "15th Five-Year Plan" explicitly identifies embodied intelligence as a new driver of economic growth.
The Yangtze River Delta boasts a strong manufacturing base. Combined with the benefits of Jiangsu Province's series of "AI+" policies, Jiangsu has a uniquely favorable industrial environment for developing embodied intelligent agents for hardware terminals. Although the current industry boasts abundant open-source software, models and data lack a bridge to hardware-side implementation, which limits the mass production and application of intelligent agents in physical terminals.
At the same time, the OPC model is making the transition from software OPC to hardware OPC. Software OPC is "human + AI tools" to quickly complete software validation, while hardware OPC is "human + AI + equipment supply chain," which requires completing sensor adaptation, edge-side model embedded development, prototype manufacturing, and real-world scenario testing and delivery. Hardware OPC innovation is more challenging and urgently requires the integration of "models, chips, cloud, and applications" to form "system-integrated" intelligent terminal applications.
In response to this industry bottleneck, Speechzen has taken the lead in establishing China's first open-source community focused on the vertical field of "hardware terminals + intelligent agents" – the Embodied Intelligent Agent Community (hereinafter referred to as the "EIA Community"). This initiative aims to build an integrated, sustainable, and innovative open-source ecosystem encompassing "models, chips, cloud, and applications," thereby addressing the shortcomings in the implementation of edge-side intelligence. The Embodied Intelligence Community focuses on mass-producible edge-side intelligent product solutions, adhering to the principles of edge-cloud collaboration, hardware-software integration, industry orientation, and independent controllability. It performs in-depth adaptation for domestic chips and operating systems, bridging the critical gap from algorithm prototypes to the implementation of hardware products.
At the meeting, Yu Kai focused on introducing the Embodied Intelligence Community's Open-Bench platform, which is an open, reproducible AI experimental evaluation platform. The platform focuses on embodied intelligence that combines hardware and software, as well as open-source model evaluation scenarios. It can solve challenges such as protocol-driven fair evaluation, paper reproduction within a unified framework, composable agent orchestration, and end-to-cloud integrated online simulation. This establishes a complete closed loop for agent implementation, from cloud-based evaluation to online simulation of virtual physical scenarios, forming a sustainable iterative feedback loop. It helps research teams and small and medium-sized enterprises reduce costs related to hardware-software compatibility and prototype development, significantly shortening the R&D cycle for intelligent terminal products. This facilitates the transition of solutions like embodied robotics and intelligent manufacturing from proof of concept to real-world business deployment.
During the conference, Shen Jianrong, Deputy Governor of Jiangsu Province, and his delegation visited the Speechow exhibition area. Yu Kai briefed Shen Jianrong and his delegation on the development of Speechow's embodied intelligence open-source community and the innovative achievements in areas such as smart office and smart IoT.
The Smart Office products exhibited by Speechow at this event include the Ceiling Microphone Series, AI Office Notebook Series, AI Camera, Recording Card, AI Super Pen, and other hardware products.
Among them, the ceiling microphone series, designed for government and corporate offices and paired with AI cameras, can create a clear audio-visual experience for both in-person and remote work, addressing issues such as unclear sound pickup, significant reverberation, and difficulty in seeing whiteboard content during meetings. The AI Office Notebook series, AI Recording Card, and AI Super Pen, which are designed for individual users, primarily address common challenges encountered in personal office work. The AI Notebook is designed for in-depth meeting scenarios, supports AI insights, and can provide action recommendations after the meeting, achieving a functional leap from "recording" to "decision-making." The AI Recording Card features an ultra-slim and highly portable design, serving as a pickup solution for mobile and fragmented scenarios. The AI Super Pen supports both handwriting and recording, allowing users to quickly locate corresponding audio by selecting handwritten notes on paper. These three products cover common personal office scenarios such as meetings and discussions, mobile data collection, and handwritten notes.
In the field of smart IoT, Speechzen showcased its embodied robot solution — the Tianji Fusion multimodal large model interaction system. This system enables robots to go beyond simply "understanding speech." They can now comprehend their environment, recognize gestures and emotions, think, and execute appropriate actions. This marks a true leap from "being able to hear and speak" to "being able to collaborate and understand emotions." The system can be adapted to various forms, including humanoid robots, service robots, robot vacuum cleaners, retail robots, and showroom guide robots.
In the future, Speech будет rely on the open-source platform of the Intelligent Community to drive the deep empowerment of the real economy by AI technology, facilitate the widespread adoption of the next generation of intelligent terminals and intelligent agents, and continue to deliver open-source innovation for the high-quality development of the artificial intelligence industry in Jiangsu and across the country.
