Machine Learning for Social Robotics in Education: Adapting RoboLearnPI's Behavior to Student Interactions

V. Mathivanan; Vimal Kumar Stephen; Ramesh Palanisamy; Senthil Jayapal; Mohammed Tauqeer Ullah; Mohamed R. Rafi

doi:10.3991/ijim.v20i12.62168

Authors

V. Mathivanan University of Technology and Applied Sciences, Ibra, Sultanate of Oman
Vimal Kumar Stephen University of Technology and Applied Sciences, Ibra, Sultanate of Oman
Ramesh Palanisamy University of Technology and Applied Sciences, Ibra, Sultanate of Oman
Senthil Jayapal University of Technology and Applied Sciences, Ibra, Sultanate of Oman
Mohammed Tauqeer Ullah University of Technology and Applied Sciences, Ibra, Sultanate of Oman
Mohamed R. Rafi University of Technology and Applied Sciences, Ibra, Sultanate of Oman

DOI:

https://doi.org/10.3991/ijim.v20i12.62168

Keywords:

Social Robotics in Education, RoboLearnPI behavior, Students interaction, Learning outcome, Reinforcement Learning, Social behavior

Abstract

A potential opportunity of social robotics integration into education is a personalization of learning in which the behavior of robots can adapt dynamically to the needs of a specific student. In this paper, an innovative framework named ACBSO-SR-RoboLearnPI-A2C is suggested and will be based on Adaptive Chaos Bird Swarm Optimization (ACBSO) along with an Asynchronous Actor-Critic (A2C) reinforcement learning model to improve adaptive behavior and interaction in educational robots. This framework uses real-time interaction data, such as response times, correctness, and engagement indicators, of the Omani Higher Education Student Performance Dataset (Moodle + SIS + eDify) to control the adaptive instructional and social behavior of the RoboLearnPI robot. The suggested method is compared with the traditional baselines such as Deep Q-Network (DQN), Proximal Policy Optimization (PPO), standard A2C, and a non-adaptive model. The quantitative outcomes show that the ACBSOSR-RoboLearnPI-A2C framework has better performance with various measurements: an Accuracy of 91.8, Learning Gain of 19.8, student engagement of 86, and low prediction errors (MAE = 0.214, RMSE = 0.296, MAPE = 7.6%). Convergence analysis indicates steadier and quicker policy learning, whereas ablation studies indicate the importance of ACBSO, chaos mechanism, and engagement-based reward shaping to overall performance. The reliability of these improvements is statistically validated (paired t-test, p < 0.05). Qualitative analysis also shows an increase in the responsiveness of social and adaptive patterns of interaction. The findings of this paper demonstrate the usefulness of combining the ACBSO-based adaptive learning in learning social robotics to enhance student learning. On the whole, the provided research creates a sound paradigm of adaptive intelligent social robotics in education, which shows considerable enhancement in the student learning outcomes, interaction, and engagement.

References

[1] Liu, D., Huang, R., Chen, Y., Adarkwah, M. A., Zhang, X., Li, X., Zhang, J., & Da, T. (2024). Using educational robots to enhance learning: An analysis of 100 academic articles (Smart Computing and Intelligence). Springer Singapore. https://doi.org/10.1007/978-981-97-5826-5

[2] Lampropoulos, G. (2025). Social Robots in Education: Current Trends and Future Perspectives. Information, 16(1), 29. https://doi.org/10.3390/info16010029

[3] Maj, K., Gołębicka, A., & Siwińska, Z. (2025). How children learn from robots: Educational implications of communicative style and gender in child–robot interaction. Computers & Education, 239, 105445. https://doi.org/10.1016/j.compedu.2025.105445

[4] Chu, S.-T., Hwang, G.-J., & Tu, Y.-F. (2022). Artificial intelligence-based robots in education: A systematic review of selected SSCI publications. Computers and Education: Artificial Intelligence, 3, 100091. https://doi.org/10.1016/j.caeai.2022.100091

[5] Akalin, N., & Loutfi, A. (2021). Reinforcement Learning Approaches in Social Robotics. Sensors, 21(4), 1292. https://doi.org/10.3390/s21041292

[6] Ackermann, H., Lange, A.L., Hafner, V.V. et al. How adaptive social robots influence cognitive, emotional, and self-regulated learning. Sci Rep 15, 6581 (2025). https://doi.org/10.1038/s41598-025-91236-0

[7] Tozadore, D., Romero, R. Student Behaviour Modelling and Adaptive Techniques for Social Robots: Data-driven and Teacher-Perceived Evaluations. Int J of Soc Robotics (2025). https://doi.org/10.1007/s12369-025-01326-2

[8] Youssef, K., Said, S., Alkork, S., & Beyrouthy, T. (2023). Social Robotics in Education: A Survey on Recent Studies and Applications. International Journal of Emerging Technologies in Learning (iJET), 18(03), 67–82. https://doi.org/10.3991/ijet.v18i03.33529

[9] Tosic, D., & et al. (2017). Robot Online Learning to Lift Weights: A Way to Expose Students to Reinforcement Learning. International Journal of Online and Biomedical Engineering (iJOE), 13(08), 4–19. https://doi.org/10.3991/ijoe.v13i08.7270

[10] Alajmi, A., et al. (2026). Learner Behavior Recognition and Dynamic Guidance Strategy Based on Reinforcement Learning in Online Education. International Journal of Interactive Mobile Technologies (iJIM) DOI: https://doi.org/10.3991/ijim.v20i06.60865

[11] Karalekas, G., Vologiannidis, S., & Kalomiros, J. (2025). Teaching Artificial Intelligence and Machine Learning in Secondary Education: A Robotics-Based Approach. Applied Sciences, 15(8), 4570. https://doi.org/10.3390/app15084570

[12] Karalekas, G., Vologiannidis, S., & Kalomiros, J. (2023). Teaching Machine Learning in K–12 Using Robotics. Education Sciences, 13(1), 67. https://doi.org/10.3390/educsci13010067

[13] Romero-C. de Vaca, A. J., Melendez-Armenta, R. A., & Ponce, H. (2024). Using Social Robotics to Identify Educational Behavior: A Survey. Electronics, 13(19), 3956. https://doi.org/10.3390/electronics13193956

[14] Alam, A. (2022). Social robots in education for long-term human–robot interaction: Socially supportive behaviour of robotic tutor for creating robo-tangible learning environment in a guided discovery learning interaction. ECS Transactions, 107(1), 12389–12403. https://doi.org/10.1149/10701.12389ecst

[15] Love, R., Cohen, P. R., Venture, G., & Kulić, D. (2025). Adapting a teachable robot’s dialog responses using reinforcement learning: Cross-cultural user study exploring effect on engagement. ACM Transactions on Human-Robot Interaction, 14(4), Article 65, 1–34. https://doi.org/10.1145/3743692

[16] Ntomora, P.-I., & Petrakos, K. (2024). Leveraging social network analysis in education through social robots: A review. Global Journal of Engineering and Technology Advances, 20(1), 55–66. https://doi.org/10.30574/gjeta.2024.20.1.0113

[17] de Greeff, J., & Belpaeme, T. (2015). Why robots should be social: Enhancing machine learning through social human–robot interaction. PLoS ONE, 10(9), e0138061. https://doi.org/10.1371/journal.pone.0138061

[18] Dong, J., Mohd Rum, S. N., Kasmiran, K. A., Mohd Aris, T. N., & Mohamed, R. (2022). Artificial intelligence in adaptive and intelligent educational system: A review. Future Internet, 14(9), 245. https://doi.org/10.3390/fi14090245

[19] Gligorea, I., Cioca, M., Oancea, R., Gorski, A. T., Gorski, H., & Tudorache, P. (2023). Adaptive learning using artificial intelligence in e-learning: A literature review. Education Sciences, 13(12), 1216. https://doi.org/10.3390/educsci13121216