Abstract
As the locomotion capabilities of quadruped robots are further developed, they are increasingly deployed in more complex environments, heightening the risk of motor failures and significantly impacting their performance. Autonomous adaptation to such failures is crucial to ensure continued operation or safe return. In this paper, we propose a learning-based framework that enables quadruped robots to adapt to single-motor failure and maintain reliable locomotion. Our approach combines a teacher-student framework with a reliability reward term to learn adaptive and robust control policies. The teacher network, which has access to privileged information about motor failures, guides the learning of the student network, which relies solely on a history of proprioceptive observations. The reliability reward term encourages the robot to lift the weak leg to a safe height, mitigating the risks associated with motor failures. We evaluate our framework through extensive simulation experiments, analyzing the adaptability and reliability of the learned policies. The results demonstrate that our approach effectively enhances the robot's ability to maintain stable locomotion under motor failure conditions.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Bellicoso, C.D., et al.: Advances in real-world applications for legged robots. J. Field Robot. 35, 1311–1326 (2018)
Bouman, A., et al.: Autonomous spot: long-range autonomous exploration of extreme environments with legged locomotion. In: Proceedings of the 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 2518–2525 (2020)
Hooks, J., et al.: ALPHRED: a multi-modal operations quadruped robot for package delivery applications. IEEE Robot. Autom. Lett. 5, 5409–5416 (2020)
Cui, J., Li, Z., Qiu, J., Li, T.: Fault-tolerant motion planning and generation of quadruped robots synthesised by posture optimization and whole body control. Complex Intell. Syst. 8, 2991–3003 (2022)
Chen, Z., Xi, Q., Gao, F., Zhao, Y.: Fault-tolerant gait design for quadruped robots with one locked leg using the GF set theory. Mech. Mach. Theory 178, 105069 (2022)
Chen, Z., Xi, Q., Qi, C., Chen, X., Gao, Y., Gao, F.: Fault-tolerant gait design for quadruped robots with two locked legs using the GF set theory. Mech. Mach. Theory 195, 105592 (2024)
Lin, C.-M., Chen, C.-H.: Robust fault-tolerant control for a biped robot using a recurrent cerebellar model articulation controller. IEEE Trans. Syst. Man Cybern. Part B Cybern. 37, 110–123 (2007)
Kim, M., Shin, U., Kim, J.-Y.: Learning quadrupedal locomotion with impaired joints using random joint masking. arXiv:2403.00398 (2024)
Luo, Z., Xiao, E., Lu, P.: FT-Net: learning failure recovery and fault-tolerant locomotion for quadruped robots. IEEE Robot. Autom. Lett. 8, 8414–8421 (2023)
Hou, T., Tu, J., Gao, X., Dong, Z., Zhai, P., Zhang, L.: Multi-task learning of active fault-tolerant controller for leg failures in quadruped robots. arXiv:2402.08996 (2024)
Liu, D., Zhang, T., Yin, J., See, S.: Saving the limping: fault-tolerant quadruped locomotion via reinforcement learning. arXiv:2210.00474 (2023)
Wu, X., Dong, W., Lai, H., Yu, Y., Wen, Y.: Adaptive control strategy for quadruped robots in actuator degradation scenarios. In: Proceedings of the Fifth International Conference on Distributed Artificial Intelligence, pp. 1–13. Association for Computing Machinery, New York (2023)
Margolis, G.B., Yang, G., Paigwar, K., Chen, T., Agrawal, P.: Rapid locomotion via reinforcement learning. arXiv:2205.02824 (2022)
Jin, Y., Liu, X., Shao, Y., Wang, H., Yang, W.: High-speed quadrupedal locomotion by imitation-relaxation reinforcement learning. Nat. Mach. Intell. 4, 1198–1208 (2022)
Lee, J., Hwangbo, J., Wellhausen, L., Koltun, V., Hutter, M.: Learning quadrupedal locomotion over challenging terrain. Sci. Robot. 5, eabc5986 (2020)
Rudin, N., Hoeller, D., Reist, P., Hutter, M.: Learning to walk in minutes using massively parallel deep reinforcement learning. Presented at the 5th Annual Conference on Robot Learning, 19 June 2021
Wu, J., Xin, G., Qi, C., Xue, Y.: Learning robust and agile legged locomotion using adversarial motion priors. IEEE Robot. Autom. Lett. 8, 4975–4982 (2023)
Choi, S., Ji, G., Park, J., Kim, H., Mun, J., Lee, J.H., Hwangbo, J.: Learning quadrupedal locomotion on deformable terrain. Sci. Robot. 8, eade2256 (2023)
Margolis, G.B., Agrawal, P.: Walk these ways: tuning robot control for generalization with multiplicity of behavior. In: Proceedings of the 6th Conference on Robot Learning, pp. 22–31. PMLR (2022)
Wu, J., Xue, Y., Qi, C.: Learning multiple gaits within latent space for quadruped robots. arXiv:2308.03014 (2023)
He, Z., Lei, K., Ze, Y., Sreenath, K., Li, Z., Xu, H.: Learning visual quadrupedal loco-manipulation from demonstrations. arXiv:2403.20328 (2024)
Kumar, A., Fu, Z., Pathak, D., Malik, J.: RMA: rapid motor adaptation for legged robots. In: Robotics: Science and Systems XVII. Robotics: Science and Systems Foundation (2021)
Tobin, J., Fong, R., Ray, A., Schneider, J., Zaremba, W., Abbeel, P.: Domain randomization for transferring deep neural networks from simulation to the real world. In: Proceedings of the 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 23–30 (2017)
Pinto, L., Andrychowicz, M., Welinder, P., Zaremba, W., Abbeel, P.: Asymmetric actor critic for image-based robot learning. In: Robotics: Science and Systems (2018)
Chen, D., Zhou, B., Koltun, V., Krähenbühl, P.: Learning by cheating. In: Proceedings of the Conference on Robot Learning, pp. 66–75. PMLR (2020)
Schulman, J., Wolski, F., Dhariwal, P., Radford, A., Klimov, O.: Proximal policy optimization algorithms. arXiv:1707.06347 (2017)
Ross, S., Gordon, G., Bagnell, D.: A reduction of imitation learning and structured prediction to no-regret online learning. In: Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, JMLR Workshop and Conference Proceedings, pp. 627–635 (2011)
Makoviychuk, V., et al.: Isaac Gym: high performance GPU-based physics simulation for robot learning. arXiv:2108.10470 (2021)
Acknowledgement
This work was supported by the National Natural Science Foundation of China (Grant No. 52375014).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2025 The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
About this paper
Cite this paper
Wang, S., Zhao, C., Qian, L., Luo, X. (2025). Learning Fault-Tolerant Quadruped Locomotion with Unknown Motor Failure Using Reliability Reward. In: Lan, X., Mei, X., Jiang, C., Zhao, F., Tian, Z. (eds) Intelligent Robotics and Applications. ICIRA 2024. Lecture Notes in Computer Science(), vol 15208. Springer, Singapore. https://doi.org/10.1007/978-981-96-0783-9_10
Download citation
DOI: https://doi.org/10.1007/978-981-96-0783-9_10
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-96-0782-2
Online ISBN: 978-981-96-0783-9
eBook Packages: Computer ScienceComputer Science (R0)