Evolution directions and development pathways of embodied artificial intelligence

doi:10.12267/j.issn.2096-5931.2026.09.005

Abstract

Abstract:

With the deep integration of artificial intelligence (AI) and robotics, embodied AI is evolving from monolithic and closed task execution toward autonomous collaboration and continuous evolution in real dynamic environments. This paper provides an analysis of the future evolution of embodied AI, identifying heterogeneity, generalization, and distributed collaboration as key evolutionary paths. Based on these insights, this paper proposes a systematic evolution architecture comprising embodied agents, near-source agents, cloud agents, and the internet of agents, providing a reference for the scalable, collaborative, and networked development of embodied AI.

Key words: embodied artificial intelligence, robotics, agent, distributed collaboration

CLC Number:

TP242
TP18

ZHANG Wenjing, CHAO Junxian, WU Zijian, CHENG Baoping. Evolution directions and development pathways of embodied artificial intelligence[J]. Information and Communications Technology and Policy, 2026, 52(9): 38-44.

Add to citation manager EndNote|Ris|BibTeX

URL:

http://ictp.caict.ac.cn/EN/10.12267/j.issn.2096-5931.2026.09.005

http://ictp.caict.ac.cn/EN/Y2026/V52/I9/38

对比维度	能耗/J	算力/GOPS	算法	数据量/KB/s	用时/s	成功率
机器人	50	100 000	>10种(感知、识别、规划、运控等)	102 400	10	约50%(需大样本仿真训练,鲁棒性差)
自然人	0.1	20(等效)	脊髓反射+小脑自动校准	10	1	约100%(且具备容错性)
结论	机器人的能耗是自然人的500倍	机器人所需算力是自然人的5 000倍	机器人依赖大量复杂算法	数据量是自然人的10 000倍	机器人的用时是自然人的10倍	机器人的成功率低且鲁棒性较差

对比维度	能耗/J	算力/GOPS	算法	数据量/KB/s	用时/s	成功率
机器人	50	100 000	>10种(感知、识别、规划、运控等)	102 400	10	约50%(需大样本仿真训练,鲁棒性差)
自然人	0.1	20(等效)	脊髓反射+小脑自动校准	10	1	约100%(且具备容错性)
结论	机器人的能耗是自然人的500倍	机器人所需算力是自然人的5 000倍	机器人依赖大量复杂算法	数据量是自然人的10 000倍	机器人的用时是自然人的10倍	机器人的成功率低且鲁棒性较差

场景	应用类别	反应时延 /ms	数据规模 /MB/s	网络需求 /Mbit/s	算力需求 /TOPS	说明
行走与奔跑	快反应类	1~10	1~10	1~10	10~50	高实时性要求,用于运动学计算和平衡控制
目标追踪	快反应类	10~100	10~100	10~100	5~30	实时目标检测(如YOLO)和追踪算法(如DeepSORT)需要较高算力
语音识别	狭义AI类	100~500	1~10	110	2~15	实时处理语音数据,运行语音识别模型
儿童陪伴	狭义AI类	100~500	10~50	10~50	5~100	处理语音和图像数据,运行交互算法(如语音识别、情感识别)
自适应行为学习	慢思考类	500~2 000	100~1 024	10~100	50~275	处理大量历史数据,运行机器学习模型(如强化学习、深度学习、创造创新、推理、自我意识)
通用AI	慢思考类	10~500	1 024~ 10 240	100~ 10 240	100~1 000	实时处理多模态数据(如视觉、语音、文本),运行复杂的AI模型和决策算法,具备类似人类的智能水平,能够处理多种复杂任务并适应动态环境

场景	应用类别	反应时延 /ms	数据规模 /MB/s	网络需求 /Mbit/s	算力需求 /TOPS	说明
行走与奔跑	快反应类	1~10	1~10	1~10	10~50	高实时性要求,用于运动学计算和平衡控制
目标追踪	快反应类	10~100	10~100	10~100	5~30	实时目标检测(如YOLO)和追踪算法(如DeepSORT)需要较高算力
语音识别	狭义AI类	100~500	1~10	110	2~15	实时处理语音数据,运行语音识别模型
儿童陪伴	狭义AI类	100~500	10~50	10~50	5~100	处理语音和图像数据,运行交互算法(如语音识别、情感识别)
自适应行为学习	慢思考类	500~2 000	100~1 024	10~100	50~275	处理大量历史数据,运行机器学习模型(如强化学习、深度学习、创造创新、推理、自我意识)
通用AI	慢思考类	10~500	1 024~ 10 240	100~ 10 240	100~1 000	实时处理多模态数据(如视觉、语音、文本),运行复杂的AI模型和决策算法,具备类似人类的智能水平,能够处理多种复杂任务并适应动态环境

[1]	梅傲然, 王健华, 张华祥, 等. 大语言模型在机器人任务规划中的应用与展望[J]. 复旦学报(自然科学版), 2026, 65(3): 345-364.
[2]	Xu Z, Wu K, Wen J, et al. A survey on robotics with foundation models: toward embodied AI[PP/OL]. arXiv(2024-02-04)[2026-08-20]. https://arxiv.org/abs/2402.02385.
[3]	Wei P, Feng W, Chen Y, et al. Robot-oriented 6G satellite-UAV networks: Requirements, paradigm shifts, and case studies[J]. China Communications, 2024, 21(2): 74-84. doi: 10.23919/JCC.fa.2023-0635.202402
[4]	胡梅, 朱学涵, 高宏玲. 具身智能机器人产业发展前瞻及建议[J]. 机器人产业, 2026 (3): 24-30.
[5]	王耀南. 具身智能与机器人技术应用及展望[J]. 中国工业和信息化, 2026 (6): 38-41.
[6]	徐俊, 郭喆晨, 王凯龙, 等. 具身智能机器人技术演进与未来展望[J/OL]. 西安交通大学学报, 2026-07-14. https://link.cnki.net/urlid/61.1069.T.20260714.1048.002.

Evolution directions and development pathways of embodied artificial intelligence

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 3

References 6

Related Articles 15

Recommended Articles

Metrics

Comments

[1]	WANG Yanmin, FENG Wei, DONG Chao, REN Baoquan, DUAN Xiangyang. Sensing-communication-computing-control closed loop for 6G unmanned operations: model, metric, and network design [J]. Information and Communications Technology and Policy, 2026, 52(9): 2-8.
[2]	LIU Jun, DONG Ji, WANG Wei. Research on security protection and governance paths of terminal agents from perspective of risk management and control [J]. Information and Communications Technology and Policy, 2026, 52(8): 18-22.
[3]	WU Hanbing, FENG Yizhuo. Research on user authorization mechanisms and identity forgery risks in agent interconnection scenarios [J]. Information and Communications Technology and Policy, 2026, 52(8): 2-10.
[4]	HAN Han, WANG Jingyao, CHANG Tianen. A novel agent security software testing scheme driven by large language model [J]. Information and Communications Technology and Policy, 2026, 52(8): 74-81.
[5]	XU Ming. A review of cross-domain data security governance for low-altitude one-network unified flight platform [J]. Information and Communications Technology and Policy, 2026, 52(6): 70-78.
[6]	WANG Shiji, WANG Liwen, ZHANG Yan, CAO Chang, WANG Runyan, HE Wanxian, PEI Pei. Service semantic routing technology for application-level computing power services [J]. Information and Communications Technology and Policy, 2026, 52(5): 80-88.
[7]	CHENG Ying, LIU Zhipeng, ZHANG Meng. Research on development pathways, potential impacts, and governance strategies of terminal intelligent agents [J]. Information and Communications Technology and Policy, 2026, 52(3): 2-7.
[8]	YANG Lei, WANG Feng, ZHANG Xin, LIN Jun. The ecological game and comparison of cooperative modes between intelligent terminals and internet services in the large model era [J]. Information and Communications Technology and Policy, 2026, 52(3): 8-13.
[9]	LIU Gaofeng, JIANG Yuehan, ZHANG Yuxiong. Artificial intelligence enables high-level development and high-quality construction of digital twin city [J]. Information and Communications Technology and Policy, 2025, 51(3): 74-82.
[10]	XIN Xing, WU Jun, WANG Haiyan, ZHANG Yang, GUO Wei. Research on the application scenarios and trends of AI in empowering telecom user complaint management [J]. Information and Communications Technology and Policy, 2025, 51(12): 57-62.
[11]	HUANG Fengyu, HUANG Yunxia, FU Rongrong, ZHAO Qiannan. Research on the development status of artificial intelligence terminal industry [J]. Information and Communications Technology and Policy, 2025, 51(12): 92-96.
[12]	ZHANG Xi, LI Chaozhuo, XU Nuo, ZHANG Litian. Security challenges and response mechanisms for trustworthy large language model agents [J]. Information and Communications Technology and Policy, 2025, 51(1): 33-37.
[13]	MENG Nan, ZHOU Chengsheng, ZHAO Xun. Research on generative artificial intelligence empowering cybersecurity operations with noise reduction capability [J]. Information and Communications Technology and Policy, 2024, 50(8): 24-31.
[14]	REN Xian, XUE Feng, FANG Yunfei, JIANG Zhonglin, CHEN Yong, TANG Wei. Research and practice of intelligent cockpit based on large language model and agent [J]. Information and Communications Technology and Policy, 2024, 50(12): 58-63.
[15]	LIU Yang. Analysis of multi鄄agent technology and application prospects of Industrial Internet [J]. Information and Communications Technology and Policy, 2021, 47(10): 30-.