Deep Reinforcement Learning-Based Fault-Tolerant Flight Control of a Quadcopter Using Full Rotor Thrust Vectoring After Complete Rotor Failure
DOI:
https://doi.org/10.70112/arme-2026.15.2.4359Keywords:
Deep Deterministic Policy Gradient , Reinforcement Learning, Drone Fault Tolerance, Adaptive Thrust Vectoring, Pan, Tilt RotorAbstract
This article addresses the issue of rebalancing a quadcopter drone after a single rotor fails, employing an unconventional method of thrust vectoring the remaining rotors. To determine the complex control policy for managing the thrust vectoring components, the Reinforcement Learning method was used to train a Deep Deterministic Policy Gradient (DDPG) agent. The implementation of the control policy during simulation showed that the drone not only managed to avoid crashing after a single rotor failed, but it also remained airborne while maintaining an acceptable change in attitude and altitude, albeit at the cost of the drone moving backwards continuously. The agent's capability to propose a better control policy could be further improved by selecting appropriate reward functions, which should include restricting the motion of the drone in the X- and Y-axis directions.
References
[1] G. Wild, J. Murray, and G. Baxter, “Exploring civil drone accidents and incidents to help prevent potential air disasters,” Aerospace, vol. 3, no. 3, pp. 1–11, 2016, doi: https://doi.org/10.3390/aerospace3030022.
[2] Z. R. Khavas, A. Majdi, S. Ahmadzadeh, and P. Robinette, “Human trust after drone failure: Study of the effects of drone type and failure type on human-drone trust,” in 2023 20th International Conference on Ubiquitous Robots (UR), Honolulu, Hawaii, 2023, pp. 685–692, doi: https://doi.org/10.1109/UR57808.2023.10202489.
[3] Y. C. Yeh, “Triple-triple redundant 777 primary flight computer,” in 1996 IEEE Aerospace Applications Conference Proceedings, Aspen, CO, USA, 1996, vol. 1, pp. 293–307, doi: https://doi.org/10.1109/AERO.1996.495891.
[4] C. M. Pittman, Taming Your Amygdala: Brain-Based Strategies to Quiet the Anxious Mind. Eau Claire, WI, USA: PESI Publishing, Inc., 2022.
[5] G. K. Fourlas and G. C. Karras, “A survey on fault diagnosis and fault-tolerant control methods for unmanned aerial vehicles,” Machines, vol. 9, no. 9, pp. 1–34, 2021, doi: https://doi.org/10.3390/machines9090197.
[6] H. Shraim, A. Awada, and R. Youness, “A survey on quadrotors: Configurations, modeling and identification, control, collision avoidance, fault diagnosis and tolerant control,” IEEE Aerospace and Electronic Systems Magazine, vol. 33, no. 7, pp. 14–33, 2018, doi: https://doi.org/10.1109/MAES.2018.160246.
[7] M. Saied, H. Shraim, and C. Francis, “A review on recent development of multirotor UAV fault-tolerant control systems,” IEEE Aerospace and Electronic Systems Magazine, vol. 39, no. 9, pp. 146–180, 2024, doi: https://doi.org/10.1109/MAES.2023.3327697.
[8] J. Guo, J. Qi, and C. Wu, “Robust fault diagnosis and fault-tolerant control for nonlinear quadrotor unmanned aerial vehicle system with unknown actuator faults,” International Journal of Advanced Robotic Systems, vol. 18, no. 2, 2021, doi: https://doi.org/10.1177/17298814211002.
[9] S. Mallavalli and A. Fekih, “Adaptive fault tolerant control design for actuator fault mitigation in quadrotor UAVs,” in 2018 IEEE Conference on Control Technology and Applications (CCTA), Copenhagen, Denmark, 2018, doi: https://doi.org/10.1109/CCTA.2018.8511427.
[10] S. Mallavalli and A. Fekih, “An SMC-based fault tolerant control design for a class of underactuated unmanned aerial vehicles,” in 2018 4th International Conference on Control, Automation and Robotics (ICCAR), Auckland, New Zealand, 2018, doi: https://doi.org/10.1109/ICCAR.2018.8384661.
[11] L. Xulin and Y. Guo, “Fault tolerant control of a quadrotor UAV using control allocation,” in 2018 Chinese Control and Decision Conference (CCDC), Shenyang, China, 2018, doi: https://doi.org/10.1109/CCDC.2018.8407422.
[12] W. Chung and H. Son, “Fault-tolerant control of multirotor UAVs by control variable elimination,” IEEE/ASME Transactions on Mechatronics, vol. 25, no. 5, pp. 2513–2522, 2020, doi: https://doi.org/10.1109/TMECH.2020.2982436.
[13] M. D. Dehkordi and M. Danesh, “Fuzzy control allocation of a positionable rotor quadrotor based on log-barrier optimization and propulsion system fault toleration,” IET Control Theory & Applications, vol. 18, no. 9, pp. 1176–1190, 2024, doi: https://doi.org/10.1049/cth2.12653.
[14] Z. Zaludin, “Drone fault tolerance strategy by allowing remaining rotor to pan and tilt,” International Journal of Advanced Mechatronic Systems, vol. 11, no. 2, pp. 102–110, 2024, doi: https://doi.org/10.1504/IJAMECHS.2024.139180.
[15] Z. Zaludin, “Enhancing drone performance: Using reinforcement learning for active pan and tilt rotor fault tolerance strategy,” International Journal of Advanced Mechatronic Systems, vol. 12, no. 2, pp. 132–146, 2025, doi: https://doi.org/10.1504/IJAMECHS.2025.145731.
[16] Z. Zaludin, “Fault tolerance conceptual strategy for a quadcopter drone with rotor failure,” Asian Review of Mechanical Engineering, vol. 12, no. 2, pp. 1–14, 2023.
[17] MathWorks, “Reinforcement Learning Toolbox: User’s Guide (R2024b),” accessed Dec. 27, 2024. [Online]. Available: http://www.mathworks.com.
[18] A. Nandy and M. Biswas, Reinforcement Learning: With Open AI, TensorFlow and Keras Using Python. Berkeley, CA: Apress, 2017.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Centre for Research and Innovation
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
