信息通信技术与政策

信息通信技术与政策

信息通信技术与政策 ›› 2025, Vol. 51 ›› Issue (8): 15-25.doi: 10.12267/j.issn.2096-5931.2025.08.003

专题:人工智能+ 上一篇    下一篇

“人工智能+”赋能具身智能机器人新形态及关键技术应用

Key technologies and typical applications of “AI+” empowered embodied intelligence robots

李腾达, 朱紫钰, 韩子奇   

  1. 中国移动上海产业研究院,上海 201206
  • 收稿日期:2025-07-10 出版日期:2025-08-25 发布日期:2025-09-02
  • 作者简介:
    李腾达,中国移动上海产业研究院产业数智化产品部高级产品经理、工程师,长期致力于智能制造、工业互联网、数字化转型、人工智能等领域的技术及应用研究
    朱紫钰,中国移动上海产业研究院产业数智化产品部产品经理、工程师,长期致力于智能制造、自动驾驶、数字化转型、人工智能等领域的技术及应用研究
    韩子奇,中国移动上海产业研究院产业数智化产品部高级产品经理、工程师,长期致力于智能制造、工业互联网、数字化转型、人工智能等领域的技术及应用研究

LI Tengda, ZHU Ziyu, HAN Ziqi   

  1. China Mobile Shanghai Industrial Research Institute, Shanghai 201206, China
  • Received:2025-07-10 Online:2025-08-25 Published:2025-09-02

摘要:

人工智能与机器人的深度融合正催生新一代机器人形态,其中具身智能机器人因其强调物理实体与环境交互的核心特性而成为关键发展方向。聚焦“人工智能+”赋能的具身智能机器人这一特定形态,系统综述了具身智能机器人的概念演进与发展现状,着重剖析了人工智能技术在感知、认知、决策、执行及底层数据支撑等环节带来的变革;围绕多模态感知、大语言模型与深度强化学习等核心技术,结合工业制造、医疗护理、家庭服务等场景应用,展示了“人工智能+”赋能具身智能机器人的应用成果。同时,指出了计算资源消耗、算法泛化性与鲁棒性不足等现实瓶颈,并展望了更高效模型架构、跨模态协同与多领域扩张等未来趋势,为具身智能机器人的技术创新和产业落地提供了参考。

关键词: 人工智能+, 具身智能, 多模态感知

Abstract:

The deep convergence of artificial intelligence (AI) and robots has become a decisive catalyst for the next leap in robots technology, giving rise to new forms of intelligent agents. Among these, embodied intelligent robots stand out due to their core emphasis on physical embodiment and environmental interaction. Focusing on this specific form enabled by “AI+”, this paper offers a comprehensive survey of the conceptual evolution and current development of embodied intelligence robots, highlighting how AI reshapes perception, cognition, decision-making, execution, and data foundations. By examining key technologies, namely, multimodal perception, large language models, and deep reinforcement learning, and demonstrating their deployment in industrial manufacturing, healthcare, and household services, this paper illustrates the concrete achievements of “AI+” empowered embodied intelligence robots. The paper also identifies practical bottlenecks, including high computational demands and limited algorithmic generalization and robustness, and discusses future directions such as more efficient model architectures, cross-modal synergies, and broader domain expansion. These insights aim to provide references for both technological innovation and industrial adoption of embodied intelligence robots.

Key words: AI+, embodied intelligence, multimodal perception

中图分类号: