Search

Dynamic Gait Adaptive Control Strategy for Modular Exoskeletons Based on Human-Cooperative Reinforcement Learning

Author(s)

Dexi Meng

Affiliation(s)

Shandong Experimental High School, Jinan, Shandong, China

Abstract

Exoskeleton technology demonstrates significant potential in medical rehabilitation, military enhancement, and industrial assistance by augmenting human mobility. However, adaptive control of dynamic gait remains a core challenge. This paper systematically reviews modular exoskeleton control strategies based on Human-Cooperative Reinforcement Learning (HCRL), integrating biomechanical modeling, reinforcement learning algorithms, and modular design methodologies. Research demonstrates that HCRL optimizes human-machine synergy efficiency through real-time interaction, significantly improving gait adaptability (with energy consumption reduced by 15%–20%). The modular architecture enables rapid adaptation to diverse user needs (configuration time shortened by 50%). Future studies must address real-time constraints, multi-objective optimization conflicts, and robustness in user intent recognition to advance exoskeleton technology toward intelligent and personalized development.

Keywords

Human-Cooperative Reinforcement Learning (HCRL); Modular Exoskeletons; Dynamic Gait; Adaptive Control; Biomechanics

References

[1] Herr, H. (2009). Exoskeletons and orthoses: Classification, design challenges and future directions. Journal of NeuroEngineering and Rehabilitation, 6(1), 21. [2] Tucker, M. R., et al. (2020). Robust myoelectric control via deep metric learning for upper-limb prostheses. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 28(7), 1532–1541. [3] Sutton, R. S., & Barto, A. G. (2018). Reinforcement learning: An introduction (2nd ed.). MIT Press. [4] Chen, C., et al. (2020). Human-in-the-loop reinforcement learning for personalized exoskeleton control. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 28(3), 639–648. [5] Vallery, H., et al. (2013). Compliant actuation for exoskeleton-based walking enhancement. IEEE/ASME Transactions on Mechatronics, 18(2), 639–647. [6] Ronsse, R., et al. (2011). Adaptive control of a knee exoskeleton for tremor suppression. IEEE Transactions on Biomedical Engineering, 58(6), 1781–1788. [7] Liu, M., et al. (2021). Deep reinforcement learning for dynamic gait generation in lower-limb exoskeletons. IEEE Transactions on Industrial Electronics, 68(8), 7187–7196. [8] Kober, J., et al. (2013). Reinforcement learning in robotics: A survey. International Journal of Robotics Research, 32(11), 1238–1274. [9] Zhang, J., et al. (2022). Decentralized reinforcement learning for modular lower-limb exoskeletons. IEEE Robotics and Automation Letters, 7(3), 7206–7213. [10] Duan, Y., et al. (2023). Meta-learning for rapid adaptation of exoskeleton control policies. Nature Machine Intelligence, 5(4), 321–330.