Article Details

Abstract

To address the problems of single interaction modes, insufficient personalized guidance and poor adaptability to teaching scenarios in current teaching, this study designs and implements a large model interactive system based on ROS2. Using RDK as the hardware platform, the system integrates sensors such as cameras and microphones for multimodal data acquisition. Human-computer interaction is realized through Automatic Speech Recognition (ASR), Large Model inference and text-to-speech (TTS). ROS2 ensures efficient information exchange among modules, and the dialogue content is visualized on a touch panel. The system is finally applied to student-assisted learning. Experimental results show that the system features low interaction latency and stable operation; it can provide personalized communication and interaction, effectively stimulate students' interest in learning and enhance their learning initiative, providing a novel technical solution and practical approach for tutoring in primary and secondary education.

Keywords

ROS2; Large Model; personalized learning; assisted teaching

References

1. [1] Chen Li, Tang Xueping Research on the policy orientation of promoting the innovation of "Internet plus education" in China's provinces -- based on the text analysis of relevant policies in each province [J]. Research on Audio Visual Education, 2023, 44 (4): 5-16, 25
2. [2] Ou Mei The Ministry of Education releases the White Paper on Smart Education in China. Fudan Education Forum, 2025, 23 (4): 2
3. [3] Ji Xinyan, Li Hongke, Zhao Yingqi Research on cultivating students' computational thinking through educational robots: a meta-analysis based on 30 experiments and quasi experiments [J]. China Modern Education Equipment, 2025 (24): 1-4
4. [4] Zhang Jiarui, Zhang Weiming, Bi Fenglin, etc Local Lightweight Course Teaching Intelligent Assistance System Based on IPEX-LM [J]. Journal of East China Normal University (Natural Science Edition), 2024 (5): 162-172
5. [5] Li Mingzhen, Shi Hexin, Wei Hengbing, etc Design Scheme of Multi functional Home Sweeping Robot Based on ROS2 [J]. Household Appliances, 2025, 67 (12): 50-55
6. [6] Zhou Jin, Ye Junmin, Li Chao Multimodal Learning Emotion Computing: Motivation, Framework, and Suggestions [J]. Research on Electronic Education, 2021, 42 (7): 26-32, 46
7. [7] Chen Xue, Xue Man Research on Gesture Interaction Recognition of Preschool Education Robots Based on Improved YOLOv5s Model [J]. Automation and Instrumentation, 2025 (10): 179-183188
8. [8] Yue Xiao, Gao Peng Research on Facial Micro expression Recognition Method for Elderly People Based on Lightweight Convolutional Neural Network [J]. Computer Knowledge and Technology, 2026, 22 (3): 104-106
9. [9] Yan Chunman, Zhang Xiang, Wang Qingpeng Facial expression recognition based on improved MobileNetV2 [J]. Computer Engineering and Science, 2023, 45 (6): 1071-1078
10. [10] Liu Y P，Yang X K，Qu D. Investigation of Core Regions in Self-Supervised Pre-Trained Speech Models for Automatic Speech Recognition [J].Signal, Image and Video Processing，2025，19（15）：1279-1279.
11. [11] Zhang Xuexiang, Lv Wenhao, Wen Yibo, etc Design of Speech Access Control System Based on Deep Learning [J]. Journal of Shanghai University of Engineering and Technology, 2020, 34 (3): 253-257
12. [12] Li Ying, Wu Sheng, Xu Cong, etc A review of research on semantic text similarity calculation methods [J]. Software Guide, 2024, 23 (11): 1-11
13. [13] Sun Yangyang Research on Artificial Intelligence Education Robot Assisted Primary School English Vocabulary Teaching [J]. Journal of Zhangjiakou Vocational and Technical College, 2025, 38 (3): 74-77
14. [14] Liu Qiaoping Exploration of the Application of Intelligent Robots in Junior High School Information Technology Classroom Teaching [J]. Modern Information Technology, 2019, 3 (14): 119-120

Funding

High level Talent Introduction Project (qnsyrc202409)

About the authors

Wang Ruizhong (1988-), male, Miao ethnic, from Leishan, Guizhou, lecturer, master's student, research direction: Internet of Things, embedded, artificial intelligence; Corresponding author: Wu Yongling (1976-), male, Miao ethnicity, from Huishui, Guizhou, professor, doctoral student, research direction: data mining, machine learning, deep learning; You Tao (2002-), male, Han ethnicity, from Weng'an, Guizhou, undergraduate studies, research direction: Electronic Information Engineering; Lu Junlan (2004-), female, Han ethnicity, from Zunyi, Guizhou, undergraduate studies, research direction: Electronic Information Engineering; He Yi (2003-), male, Han nationality, from Weng'an, Guizhou, undergraduate student, research direction: Electronic Information Engineering.