Advances in Ball Trajectory Prediction for Light Weight Autonomous Table Tennis Robotic Arm: A Holistic Review
DOI:
https://doi.org/10.70112/ajeat-2025.14.2.4319Keywords:
Autonomous Table-Tennis Robots, Ball Trajectory Prediction, Computer Vision, Reinforcement Learning, Deep LearningAbstract
Autonomous table-tennis robots represent a cutting-edge application of reinforcement learning, deep learning, and computer vision for dynamic, real-time tasks. Ball trajectory prediction remains a critical challenge, requiring high-speed detection, tracking, and motion forecasting under spin and aerodynamic effects. Recent advancements integrating multi-vision systems with machine learning have significantly enhanced prediction accuracy [3]. Hybrid stereo vision architectures with robot-mounted cameras achieve bounce-prediction errors of 1–3 m and sub-40 cm accuracy at close range [4]. Deep neural networks report detection accuracies exceeding 81% [5], while LSTM models effectively capture nonlinear flight dynamics influenced by spin, drag, and Magnus forces. Reinforcement learning frameworks, particularly DDPG with physics-guided reward functions, enable lightweight robots to learn striking strategies, achieving hit rates exceeding 96% [6]. Coupling these approaches with predictive modules allows proactive control decisions [7]. High-speed vision systems operating above 100 fps, combined with GPU-accelerated processing, ensure millisecond-level responsiveness [8]. CNN-based spin estimation and three-dimensional reconstruction via stereo triangulation enable continuous trajectory refinement using receding-horizon prediction. Compact robotic arms and humanoid platforms leverage model-based planning and RL-driven motion generation for coordinated play. The fusion of computer vision, deep learning, and reinforcement learning defines the state of the art in autonomous table-tennis robotics and lays the groundwork for broader dynamic robotic applications.
References
[1] H.-I. Lin, T.-C. Kuo, and C.-H. Chen, “Ball tracking and trajectory prediction for table-tennis robots,” Sensors, vol. 20, no. 2, 2020.
[2] J. Tebbe et al., “Sample-efficient reinforcement learning in robotic table tennis,” arXiv preprint arXiv:2011.03275, 2020.
[3] S. Schwarcz et al., “SPIN: A high-speed, high-resolution vision dataset for ping-pong,” arXiv preprint arXiv:1912.06640, 2019.
[4] T. Gossard, J. Krismer, A. Ziegler, and A. Zell, “Table tennis ball spin estimation with an event camera,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR) Workshops, 2024.
[5] Y. Liu, L. Sun, and X. Zhou, “Accurate real-time ball trajectory estimation with onboard stereo camera system for a humanoid ping-pong robot,” Robotics Auton. Syst., vol. 114, pp. 101–112, 2019.
[6] Q. Song, “Application and research of trajectory prediction of table tennis robot,” in Proc. SPIE Conf. Robotics Autom., vol. 13213, 2024.
[7] F. He and Y. Li, “Modeling of SPM-GRU ping-pong ball trajectory prediction incorporating YOLOv4-Tiny algorithm,” PLoS ONE, vol. 19, no. 9, e0306483, 2024.
[8] C.-H. Sun, P.-Y. Wang, and C.-J. Chang, “Binocular-vision-based trajectory prediction of spinning balls,” Sensors, vol. 25, no. 5, 2025.
[9] J. Achterhold, P. Tobuschat, H. Ma, D. Büchler, M. Mühlebach, and J. Stückler, “Black-box vs. gray-box: A case study on learning table tennis ball trajectory prediction with spin and impacts,” Proc. Mach. Learn. Res. (PMLR), vol. 211, 2023.
[10] Y. Zhu, H. Yang, and W. Zhang, “Learn to return table tennis ball using Monte-Carlo-based reinforcement learning,” IEEE Access, vol. 8, pp. 172420–172429, 2020.
[11] Maier and T. Asfour, “Model-free reinforcement learning for robotic table tennis,” IEEE Robot. Autom. Lett., vol. 6, no. 3, pp. 6027–6034, 2021.
[12] Z. Guo, X. Chen, and D. Wang, “A multi-modal table tennis robot system based on frame and event cameras,” IEEE Sensors J., vol. 23, no. 7, pp. 8759–8770, 2023.
[13] J. Wiselin and R. Rahiman, “Fixed frequency sliding mode–PI control for single-phase unipolar inverters,” Asian J. Sci. Appl. Technol., vol. 2, no. 2, pp. 19–24, 2013.
[14] S. Bhattacharya, S. Roy, M. Sen, and U. Majhi, “IoT-based smart soil health management system with remote monitoring and control,” unpublished.
[15] R. Fischer, M. Müller, and T. Asfour, “Achieving human-level competitive play in robotic table tennis via deep reinforcement learning,” IEEE Robot. Autom. Mag., vol. 31, no. 1, pp. 112–124, 2025.
[16] S. Gomez-Gonzalez, Y. Nemmour, B. Schölkopf, and J. Peters, “Reliable real-time ball tracking for robot table tennis,” Robotics, vol. 8, no. 4, art. no. 90, 2019.
[17] Y. Huang, D. Büchler, O. Koç, B. Schölkopf, and J. Peters, “Jointly learning trajectory generation and hitting point prediction in robot table tennis,” in Proc. 16th IEEE-RAS Int. Conf. Humanoid Robots (HUMANOIDS), pp. 650–655, 2016.
[18] T. Gossard, J. Tebbe, A. Ziegler, and A. Zell, “SpinDOE: A ball spin estimation method for table tennis robot,” arXiv preprint arXiv:2303.03879, 2023.
[19] P. Senthil, “Effective monitoring of temperature and humidity in real-time wireless sensor method of processor and system architecture,” Asian J. Sci. Appl. Technol., vol. 7, no. 1, pp. 36–43, 2018.
[20] H. Kim and J. Park, “Application of table tennis ball trajectory and rotation-oriented prediction algorithm using artificial intelligence,” Appl. Sci., vol. 13, no. 2, 2023.
[21] M. Riedmiller, P. van der Smagt, and T. Asfour, “Reinforcement learning with model-based feedforward inputs for robotic table tennis,” Auton. Robots, vol. 47, no. 6, pp. 1387–1403, 2023.
[22] H. Wang et al., “SpikePingpong: High-frequency spike vision-based robot learning for precise striking in table tennis game,” arXiv preprint arXiv:2506.06690, 2025.
[23] S. Bhattacharya, A. K. Nayek, A. Biswas, M. Sen, and A. Halder, “An advanced Internet of Things-based water quality monitoring architecture for sustainable aquaculture leveraging long-range wide-area network communication protocol,” ES General, Oct. 9, 2025.
[24] P. Müller, A. Zell, and T. Asfour, “Catching spinning table tennis balls in simulation with end-to-end curriculum reinforcement learning,” arXiv preprint arXiv:2504.03122, 2025.
[25] S. Sharma and J. K. Sharma, “Entanglement measure based on matrix realignment,” Asian J. Sci. Appl. Technol., vol. 7, no. 1, pp. 17–19, 2018.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2025 Centre for Research and Innovation
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
