Behavision平台

Search documents
物理AI时代来临:天娱数科如何凭借Behavision平台抢占技术制高点?
Zheng Quan Shi Bao Wang· 2025-07-20 13:29
Core Insights - The rise of "Physical AI" is becoming a new focal point in global technology competition, with significant implications for human-machine interaction [2][4] - Tianyu Shuke, a pioneer in embodied intelligence, has developed the Behavision platform, establishing a complete closed loop of "perception-reasoning-execution" for the industrialization of Physical AI [2][3] Company Developments - Tianyu Shuke's Behavision platform integrates various models, including the Tianxing large model and the "Wise Questions" 3D intelligent industry model, creating a synergistic model matrix [2] - The Tianxing model utilizes 260TB of multimodal data, enabling cross-perception reasoning capabilities, while the "Wise Questions" model employs Sim2Real simulation technology for efficient training [2] - The company has built a standardized robot ontology classification system and an atomic skill library, supporting various types of robotic actions [3] Industry Standards and Ecosystem - Tianyu Shuke is not only a technology leader but also a standard setter, with its solutions recognized in the China Academy of Information and Communications Technology's "AI Agent Industry Map 1.0" [3] - The company has established a 3D motion capture and multimodal data collection base, storing 1.2 million structured 3D motion data sets and 500,000 multimodal behavior data sets [3] - The "human-shaped robot space 6D motion capture long-range data" sets are available on national data trading platforms, providing low-cost, high-quality data services to SMEs [3] Future Outlook - The emergence of Physical AI signifies a new stage in AI development, focusing on enabling machines to understand and interact with the physical world [4] - As the industry chain continues to improve, Physical AI is expected to unlock significant potential in smart manufacturing and service robots, with Tianyu Shuke's practices serving as a key paradigm [4] - The ability to dominate the technical standards and scene implementation of Physical AI will determine the future leadership in the next generation of AI development [4]
天娱数科:以“智能体+”深耕垂直场景,抢占AI Agent商业化高地
Zheng Quan Shi Bao Wang· 2025-07-20 13:26
Core Insights - The article discusses the evolution of AI technology, highlighting the transition from language interaction to autonomous execution, positioning intelligent agents as a new engine for industrial restructuring [1][4] - Gartner predicts that by 2026, 30% of enterprise services will be completed through intelligent agents, indicating a significant shift in the competitive landscape of global AI [1] Group 1: Intelligent Agent Ecosystem - Tianyu Shuke has developed a comprehensive intelligent agent system that integrates perception, reasoning, and execution, breaking the limitations of traditional AI that only focuses on language interaction [2] - The perception layer utilizes the self-developed Behavision platform, which combines a massive 3D corpus and deep vision chip technology, enabling machines to possess human-like intuition [2] - The reasoning layer is built on proprietary large models, such as the "Tianxing" base model, tailored for various industries like marketing and manufacturing, creating specialized decision-making capabilities [2] - The execution layer allows for seamless integration of digital decisions into physical actions through a standardized robot classification system and an atomic skill library [2] Group 2: Practical Applications in Various Industries - The Behavision platform offers a universal solution that significantly reduces the cost of intelligent agent development by using a combination of real data and extensive simulation training [3] - In marketing, the intelligent agent ecosystem, centered around a smart advertising model, addresses high traffic costs and user engagement challenges, exemplified by successful campaigns in the insurance and beauty sectors [3] - The tourism sector benefits from intelligent agents that analyze user behavior data to generate personalized recommendations, enhancing operational efficiency by 150% and achieving over 99% in customer response and accuracy rates [4]
天娱数科:公司及子公司生产经营秩序正常
Zheng Quan Shi Bao Wang· 2025-07-19 07:06
Group 1 - The core point of the announcement is that the criminal detention of the company's director, Guo Baichun, is unrelated to the company and will not significantly impact its operations [1] - The company reported a forecasted net profit attributable to shareholders of between 18 million to 26 million yuan for the first half of 2025, marking a successful turnaround from losses in the same period last year [1] - The company has developed three enterprise-level large models that have been registered with the national generative artificial intelligence service, which are expected to enhance its capabilities in digital transformation across various industries [1] Group 2 - The company has made substantial progress in data accumulation, with its Behavision platform achieving over one million sets of 3D data and 350,000 sets of multimodal data, providing strong support for AI and humanoid robot technologies [2] - A strategic cooperation was established between the company's Shanxi Data Flow Valley and the Shanxi Data Trading Center to optimize data empowerment for the real economy, marking an important layout in the field of data element circulation [2]