Interaction", led by iFLYTEK and jointly completed by the University of Science and Technology of China, Peking University, and other units, won the first prize of the Anhui Science and Technology Progress Award. At almost the same time, the project of "Forward Design and Application of 350-1200kg High-performance Heavy-load Robot" jointly developed by Midea Blue Orange Laboratory (National Key Laboratory for High-end Heavy-load Robots) and KUKA Robotics (Guangdong) Co., Ltd. successfully passed the expert appraisal organized by the China Machinery Industry Federation. The overall technology reached the international advanced level, and some technologies reached the international leading level. These two achievements, one soft and one hard, seemingly belong to different tracks, but they point to the same main line: Chinese enterprises are leveraging deep collaboration between industry, academia, and research to climb up the value chain simultaneously in the two strategic fields of artificial intelligence and high-end equipment manufacturing.

Let's first look at iFLYTEK. Human-computer interaction (HCI) serves as the fundamental support in the IoT era and is also a key direction in China's "New-generation Artificial Intelligence Development Plan". The award-winning project addresses several major pain points in multimodal HCI: weak multimodal perception in complex scenarios, poor cross-modal understanding in voice conversations, insufficient expressiveness in digital human voice and image, and difficulty in customizing interaction systems. It has achieved breakthroughs in key technologies for multimodal digital human interaction. What deserves more attention is its industrialization focus: based on innovative breakthroughs in related technologies, iFLYTEK has built an efficient and scalable digital human interaction customization platform, effectively shortening the response time of digital human interaction systems. This demonstrates that the company has not remained at the "paper-level innovation" stage, but has transformed its technological strengths into platform-based and reusable engineering capabilities, enabling its achievements to be invoked with low thresholds in more scenarios. Furthermore, the project "Key Technologies and Clinical Applications of Multimodal Imaging Diagnosis for Digestive System Tumors" participated by iFLYTEK also won the first prize of the Science and Technology Progress Award, indicating that multimodal AI is steadily penetrating from interaction scenarios into deeper industries such as healthcare.

Let's look at Midea. Heavy-duty robots in the 350 to 1200 kg class are key production equipment in industries such as automotive and heavy equipment, requiring high technical thresholds and stringent reliability standards. The project that passed the appraisal achieved three major core technological breakthroughs: it is the first to integrate typical operating conditions and other design tools into the forward design process, establishing a multi-level fault tree model to achieve a high reliability index of 100,000 hours MTBF; it innovates a quality redistribution topology optimization method, achieving a leap in overall machine performance through dynamic iterative optimization; and it breaks through multi-factor rigid-flex coupling modeling technology, enabling high-precision simulation and control optimization of dynamic characteristics. The appraisal committee, composed of seven industry experts including Professor Sun Linning from Soochow University and Academician Yang Huayong from Zhejiang University, unanimously recognized that the project results possess independent intellectual property rights. The term "forward design" carries significant weight—it signifies that domestic heavy-duty robots no longer rely on reverse imitation, but instead define products independently based on demand and operating conditions. This marks a breakthrough in key technologies in the field of high-end heavy-duty robots and their core components, elevating the overall level of self-reliance and self-improvement across the entire industry chain. It is intriguing to note that KUKA, a renowned robotics company originating from Germany, has now deeply integrated into China's innovation system, becoming a footnote in the story of multinational resources nurturing local research and development.



7月下旬，两则来自中国企业的科技消息先后传来：7月20日，安徽省科技创新大会在合肥召开，由科大讯飞牵头、联合中国科学技术大学、北京大学等单位完成的“跨模态协同的数字人交互关键技术及产业化”项目，获得安徽省科学技术进步奖一等奖；几乎同一时间，美的蓝橙实验室（高端重载机器人全国重点实验室）联合库卡机器人（广东）有限公司等单位研发的“350-1200kg高性能重载机器人正向设计及应用”项目，顺利通过中国机械工业联合会组织的专家鉴定，总体技术达到国际先进水平，部分技术达到国际领先水平。一软一硬，两项成果看似分属不同赛道，背后却指向同一条主线：中国企业正以产学研深度协同为杠杆，在人工智能与高端装备制造两大战略领域同步向价值链上游攀升。

先看科大讯飞。人机交互是物联网时代的基础支撑，也是国家“新一代人工智能发展规划”的重点方向。此次获奖项目直面多模态人机交互的几大痛点——复杂场景下多模态感知能力弱、语音对话中跨模态理解能力差、数字人声音形象表现力不足、交互系统定制难——并在多模态数字人交互关键技术上实现突破。更值得关注的是其产业化落点：基于相关技术的创新突破，讯飞搭建了高效可扩展的数字人交互定制平台，有效缩短了数字人交互系统的响应时间。这说明企业没有停留在“论文级创新”，而是把技术长板转化为平台化、可复用的工程能力，让成果能被更多场景低门槛调用。此外，讯飞参与的“消化系统肿瘤多模态影像诊断关键技术及临床应用”项目同获科学技术进步奖一等奖，显示多模态人工智能正从交互场景向医疗等纵深行业稳步渗透。

再看美的。350至1200公斤级重载机器人是汽车、重型装备等行业的关键生产装备，技术门槛高、可靠性要求苛刻。此次通过鉴定的项目实现了三大核心技术突破：首次将典型工况等设计工具融入正向流程，建立多层级故障树模型，实现10万小时MTBF的高可靠性指标；创新质量再分配拓扑优化方法，通过动态迭代优化实现整机性能跃升；突破多因素刚柔耦合建模技术，实现动态特性的高精度仿真与控制优化。由浙江大学杨华勇院士担任主任委员、苏州大学孙立宁教授等七位行业专家组成的鉴定委员会，一致认可项目成果拥有自主知识产权。“正向设计”四个字分量尤重——它意味着国产重载机器人不再依赖逆向模仿，而是从需求与工况出发自主定义产品，高端重载机器人及其核心零部件领域的关键技术实现突破，全产业链自立自强水平由此整体抬升。耐人寻味的是，库卡这家源自德国的机器人名企，如今已深度嵌入中国创新体系，成为跨国资源反哺本土研发的注脚。




The commonalities of the two achievements are clearly discernible. Firstly, both are the products of deep collaboration between industry, academia, and research: iFLYTEK has partnered with universities and institutes such as the University of Science and Technology of China and Peking University, as well as industry partners such as China Mobile (Hangzhou). Midea relies on national key laboratories and collaborates with enterprises such as Kuka and Guangdong Jiya Precision Machinery. The innovation entities are no longer lone fighters, but joint fleets formed around key links in the industrial chain. Secondly, both achievements are benchmarked against authoritative third-party evaluations: one has been tested by the provincial science and technology award system, while the other has been appraised by experts from industry associations. The quality of the achievements is well-founded, avoiding self-serving claims. Thirdly, both emphasize the closed loop from key technologies to industrial applications, addressing the longstanding issue of scientific and technological achievements "sleeping in the laboratory".

Looking ahead, two clues are worth keeping track of. For iFLYTEK, whether the digital human interactive customization platform can be replicated on a large scale in scenarios such as education, government affairs, and customer service determines whether it can convert the gold content of awards into tangible business growth. For Midea, the heavy-duty robots need to undergo multiple tests of cost control, long-term stability, and market trust before moving from "passing the evaluation" to "mass production". At the industrial level, artificial intelligence and robots are accelerating their integration: combining intelligent interaction technology with cross-modal perception and understanding capabilities with highly reliable and precise execution bodies is the complete puzzle of intelligent manufacturing. These two companies happen to hold two pieces of the puzzle respectively - one related to how machines "understand and perceive", and the other related to how machines "execute stably and accurately". If Chinese enterprises can continuously strengthen the entire chain of "perception-decision-execution" and make the collaboration of industry, academia, and research move from a project-based system to a normalized mechanism, the global discourse power of "China's intelligent manufacturing" will not only rest on individual awards and evaluation certificates, but will also precipitate into systematic industrial competitiveness. These two news stories are vivid slices of this process.



两项成果的共同点清晰可辨。其一，都是产学研深度协同的产物：讯飞联手中科大、北大等高校院所以及中移（杭州）等产业伙伴，美的依托全国重点实验室并携手库卡、广东极亚精机等企业，创新主体早已不是单打独斗，而是围绕产业链关键环节组建的联合舰队。其二，都以权威第三方评价为里程碑：一个经省级科技奖励体系检验，一个经行业联合会专家鉴定，成果成色有据可依，避免了自说自话。其三，都强调从关键技术到产业化应用的闭环，回应了科技成果“沉睡在实验室”的老问题。

展望未来，两条线索值得持续跟踪。对科大讯飞而言，数字人交互定制平台能否在教育、政务、客服等场景规模化复制，决定其能否把奖项含金量兑换为实实在在的商业增长；对美的而言，重载机器人从“通过鉴定”到“批量装机”，还需经受成本控制、长期稳定性与市场信任的多重考验。放大到产业层面，人工智能与机器人正在加速融合：具备跨模态感知与理解能力的智能交互技术，与高可靠、高精度的执行本体相结合，才是智能制造的完整拼图。两家企业恰好分别握住了拼图的两块——一块关乎机器如何“听懂看懂”，一块关乎机器如何“稳准执行”。若中国企业能在“感知—决策—执行”全链条上持续补强，并让产学研协同从项目制走向常态化机制，“中国智造”的全球话语权将不只停留在个别奖项与鉴定证书上，而会沉淀为体系化的产业竞争力。这两则新闻，正是这一进程的生动切片。

(Using AI translation)

（使用AI翻译）