Abstract
Deep reinforcement learning (DRL) has recently received a lot of attention due to its better performance compared to traditional algorithms in mapless navigation tasks. However, random exploration and sparse rewards may lead to inefficient DRL training in complex scenarios. In this paper, we first propose a gap-guided controller switching strategy to significantly improve the training efficiency of DRL by introducing human-designed priori knowledge. Specifically, we design a reward-based controller switching algorithm that allows the priori controller (expert) to correct the robot behavior in case of poor DRL controller performance, which can avoid random exploration during training. In addition, a gap detection algorithm is used as an online-mapless planner to compute a passable sub-goal within a limited field of view, ensuring that the robot receives dense rewards in complex scenarios. In summary, our method incorporates prior knowledge into network training by combining online-mapless planner and expert demonstration, which greatly improves training efficiency. We apply our method to mapless navigation tasks in both simulation and real worlds to demonstrate that the proposed method improves exploration efficiency by 62%, total reward by 42%, and training speed by 60% compared to the normal DRL methods.
Similar content being viewed by others
Data Availability
The demo video of our proposed mapless navigation scheme is available in https://www.youtube.com/playlist?list=PLxpfeFCAyx62pI58wV9sDCWBTtK9xgUSq.
References
DeSouza, G.N., Kak, A.C.: Vision for mobile robot navigation: a survey. IEEE Trans. Pattern Anal. Mach. Intell. 24(2), 237–267 (2002)
Zhang, Z., Qin, J., Wang, S., Kang, Y., Liu, C.: Ulodnet: a unified lane and obstacle detection network towards drivable area understanding in autonomous navigation. J. Intell. Robot. Syst. 105(1), 1–11 (2022)
Chang, X., Pan, H., Sun, W., Gao, H.: Yoltrack: multitask learning based real-time multiobject tracking and segmentation for autonomous vehicles. IEEE Trans. Neural Netw. Learn. Syst. 32(12), 5323–5333 (2022)
Zhu, Y., Mottaghi, R., Kolve, E., Lim, J.J., Gupta, A., Fei-Fei, L., Farhadi, A.: Target-driven visual navigation in indoor scenes using deep reinforcement learning. In: 2017 IEEE International Conference on Robotics and Automation (ICRA), pp. 3357–3364 (2017)
Zhang, J., Springenberg, J.T., Boedecker, J., Burgard, W.: Deep reinforcement learning with successor features for navigation across similar environments. In: 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 2371–2378 (2017)
Lillicrap, T., Hunt, J., Pritzel, A., Heess, N., Erez, T., Tassa, Y., Silver, D., Wierstra, D.: Continuous control with deep reinforcement learning. arXiv:1509.02971 (2015)
Schulman, J., Wolski, F., Dhariwal, P., Radford, A., Klimov, O.: Proximal policy optimization algorithms. arXiv:1707.06347 (2017)
Haarnoja, T., Zhou, A., Hartikainen, K., Tucker, G., Ha, S., Tan, J., Kumar, V., Zhu, H., Gupta, A., Abbeel, P., et al.: Soft actor-critic algorithms and applications. arXiv:1812.05905 (2018)
Niroui, F., Zhang, K., Kashino, Z., Nejat, G.: Deep reinforcement learning robot for search and rescue applications: exploration in unknown cluttered environments. IEEE Robot. Autom. Lett. 4(2), 610–617 (2019)
Li, H., Zhang, Q., Zhao, D.: Deep reinforcement learning-based automatic exploration for navigation in unknown environment. IEEE Trans. Neural Netw. Learn. Syst. 31(6), 2064–2076 (2019)
Botteghi, N., Sirmacek, B., Mustafa, K., Poel, M., Stramigioli, S.: On reward shaping for mobile robot navigation: a reinforcement learning and slam based approach. arXiv:2002.04109 (2020)
Tai, L., Liu, M.: Mobile robots exploration through cnn-based reinforcement learning. Robot. Biomimetics 3(1), 1–8 (2016)
Kato, Y., Kamiyama, K., Morioka, K.: Autonomous robot navigation system with learning based on deep q-network and topological maps. In: 2017 IEEE/SICE International Symposium on System Integration (SII), pp. 1040–1046 (2017)
Faust, A., Oslund, K., Ramirez, O., Francis, A., Tapia, L., Fiser, M., Davidson, J.: Prm-rl: long-range robotic navigation tasks by combining reinforcement learning and sampling-based planning. In: 2018 IEEE International Conference on Robotics and Automation (ICRA), pp. 5113–5120 (2018)
Chen, G., Pan, L., Xu, P., et al.: Robot navigation with map-based deep reinforcement learning. In: 2020 IEEE International Conference on Networking, Sensing and Control (ICNSC), pp. 1-6 (2020)
Argall, B., Chernova, S., Veloso, M., Browning, B.: A survey of robot learning from demonstration. Robot. Auton. Syst. 57(5), 469–483 (2009)
Wu, Q., Gong, X., Xu, K., Manocha, D., Dong, J., Wang, J.: Towards target-driven visual navigation in indoor scenes via generative imitation learning. IEEE Robot. Autom. Lett. 6(1), 175–182 (2021)
Rana, K., Dasagi, V., Talbot, B., Milford, M., Sünderhauf, N.: Multiplicative controller fusion: leveraging algorithmic priors for sample-efficient reinforcement learning and safe sim-to-real transfer. In: 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 6069–6076 (2020)
Bellegarda, G., Byl, K.: An online training method for augmenting mpc with deep reinforcement learning. In: 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 5453–5459 (2020)
Khatib, O.: Real-time obstacle avoidance for manipulators and mobile robots. In: 1985 IEEE International Conference on Robotics and Automation (ICRA), pp. 500–505 (1985)
Fox, D., Burgard, W., Thrun, S.: The dynamic window approach to collision avoidance. IEEE Robot. Autom. Mag. 4(1), 23-33 (1997)
Ballesteros, J., Urdiales, C., Velasco, A., Ramos-Jiménez, G.: A biomimetical dynamic window approach to navigation for collaborative control. IEEE Trans. Hum.-Mach. Syst. vol. 47, no. 6, pp. 1123–1133, 2017
Khatib, O.: Real-time obstacle avoidance for manipulators and mobile robots. In: Autonomous Robot Vehicles, pp. 396–404 (1986)
Gao, Z., Qin, J., Wang, S., Wang, Y.: Boundary gap based reactive navigation in unknown environments. IEEE/CAA J. Autom. Sinica 8(2), 468–477 (2021)
Qiu, J., Ma, M., Wang, T., Gao, H.: Gradient descent-based adaptive learning control for autonomous underwater vehicles with unknown uncertainties. IEEE Trans. Neural Netw. Learn. Syst. 32(12), 5266–5273 (2021)
Li, Y., Qin, F., Du, S., Xu, D., Zhang, J.: Vision-based imitation learning of needle reaching skill for robotic precision manipulation. J. Intell. Robot. Syst. 101(1), 1–13 (2021)
Yan, C., Qin, J., Liu, Q., Ma, Q., Kang, Y.: Mapless navigation with safety-enhanced imitation learning. IEEE Trans. Ind. Electron. in press (2022)
Yan, C., Xiang, X., Wang, C.: Towards real-time path planning through deep reinforcement learning for a uav in dynamic environments. J. Intell. Robot. Syst. 98(2), 297–309 (2020)
Mujahad, M., Fischer, D., Mertsching, B., Jaddu, H.: Closest gap based (cg) reactive obstacle avoidance navigation for highly cluttered environments. In: 2010 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 1805–1812 (2010)
Thrun, S.: Probabilistic robotics. Commun. ACM 45(3), 52–57 (2002)
Grisetti, G., Stachniss, C., Burgard, W.: Improving grid-based slam with rao-blackwellized particle filters by adaptive proposals and selective resampling. In: 2005 IEEE International Conference on Robotics and Automation (ICRA), pp. 2432–2437 (2005)
Acknowledgements
We acknowledge the support of the GPU cluster built by MCCLab of Information Science and Technology Institution, USTC.
Funding
This work is supported by the Science and Technology Major Project of Anhui Province (Grant No. 202203a06020011), and by the Mobility Programme 2021 of Sino-German Center for Research Promotion (No. M-0582), and by the Open Fund of the Laboratory of Cognition and Decision Intelligence for Complex Systems, Institute of Automation, Chinese Academy of Sciences (Grant No. CASIA-KFKT-14). Part of the work of Q. Liu was carried out when he was a research fellow at the Chair of Information-Oriented Control, Technical University of Germany, Germany. (Corresponding Author: Jiahu Qin)
Author information
Authors and Affiliations
Contributions
Heng Li: Coding and writing; Jiahu Qin: Writing and review; Qingchen Liu: Writing and review; Chengzhen Yan: Review.
Corresponding author
Ethics declarations
Ethical Approval
Not applicable.
Consent to Participate
Not applicable.
Consent for Publication
Not applicable.
Conflicts of interest
The authors have no relevant financial or non-financial interests to disclose.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Li, H., Qin, J., Liu, Q. et al. An Efficient Deep Reinforcement Learning Algorithm for Mapless Navigation with Gap-Guided Switching Strategy. J Intell Robot Syst 108, 43 (2023). https://doi.org/10.1007/s10846-023-01888-1
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s10846-023-01888-1