GFlowLoc: GFlowNet-Driven Adaptive Fusion for Robust Positioning in GNSS-Challenged Environments

Azimbek Khudoyberdiev and Jihoon Ryoo

Peer Reviewed

Abstract: Reliable positioning in GNSS-challenged environments remains difficult because urban GNSS errors are structured and location-dependent. Non-line-of-sight (NLOS) and multipath propagation bias pseudorange measurements, while inertial and odometry propagation can carry positioning errors forward. Real-world driving logs capture only a limited subset of possible sensor-failure combinations. We present GFlowLoc, a GFlowNet-augmented, modality-aware fusion framework that expands training coverage through diverse, physically constrained degradation trajectories. The GFlowNet generator learns to sample degradation trajectories according to a reward combining localization difficulty and novelty, covering GNSS bias, outages, inertial drift, and odometry inconsistency. A modality-aware Mixture-of-Experts model processes GNSS, IMU, and odometry through modality-specific encoders, with adaptive gating assigning context-dependent weights to predict position corrections. A lightweight multi-hypothesis inference head combines corrections from plausible degradation hypotheses using uncertainty-aware weights. Across multi-city traces spanning open-sky roads, dense urban corridors, and long tunnels, GFlowLoc reduces the Horizontal Score by 30%–43% relative to the best external baseline for each dataset in challenging unseen environments. It also maintains stable corrections during GNSS-denied segments where deep regression baselines incur large errors. System profiling reports 0.06 ms inference latency and 0.86 GB memory usage, supporting high-rate positioning.
Published in: Proceedings of the 39th International Technical Meeting of the Satellite Division of the Institute of Navigation (ION GNSS+ 2026)
September 14 - 18, 2026
Hyatt Regency Grand Cypress
Orlando, Florida
Pages: 2723 - 2737
Cite this article: Khudoyberdiev, Azimbek, Ryoo, Jihoon, "GFlowLoc: GFlowNet-Driven Adaptive Fusion for Robust Positioning in GNSS-Challenged Environments," Proceedings of the 39th International Technical Meeting of the Satellite Division of the Institute of Navigation (ION GNSS+ 2026), Orlando, Florida, September 2026, pp. 2723-2737. https://doi.org/10.33012/2026.20756
Full Paper:

Sign in to download with your ION membership, download credits or subscription, or buy this article.

Sign In Buy this article: $12 for ION members, $35 for non-members
References:
Show 44 references
  1. Bengio, Y., Lahlou, S., Deleu, T., Hu, E. J., Tiwari, M., & Bengio, E. (2023). GFlowNet foundations. Journal of Machine Learning Research, 24(210), 1-55. https://www.jmlr.org/papers/v24/22-0364.html https://doi.org/10.1561/2200000006
  2. Evmorfos, S., Xu, Z., & Petropulu, A. (2024). Sensor selection via GFlowNets: A deep generative modeling framework to navigate combinatorial complexity. arXiv. https://doi.org/10.48550/arXiv.2407.19736
  3. Fu, G. M., Khider, M., & van Diggelen, F. (2020). Android raw GNSS measurement datasets for precise positioning. Proceedings of the 33rd International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2020), 1925-1937. https://doi.org/10.33012/2020.17628
  4. Fu, M., Khider, M., & van Diggelen, F. (2022). Summary and legacy of the Smartphone Decimeter Challenge (SDC) 2022. Proceedings of the 35th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2022), 2301-2320. https://doi.org/10.33012/2022.18379
  5. Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735-1780. https://doi.org/10.1162/neco.1997.9.8.1735
  6. Huang, X., Li, W., Dai, Z., & Zhu, X. (2025). Improving smartphone GNSS positioning in challenging urban environments using GA-BPNN. GPS Solutions, 29(1), Article 3. https://doi.org/10.1007/s10291-024-01756-x
  7. Ismaylovna, K. P., Khudoyberdiev, A., & Kim, H.-C. (2026). CogniMoE: End-to-end multimodal mental workload classification via on-the-fly scalogram generation and MoE gating. IEEE Sensors Journal, 26(3), 5213-5228. https://doi.org/10.1109/JSEN.2025.3640827
  8. Iyer, K., Dey, A., Xu, B., Sharma, N., & Hsu, L.-T. (2024). Enhancing positioning in GNSS denied environments based on an extended Kalman filter using past GNSS measurements and IMU. IEEE Transactions on Vehicular Technology, 73(6), 7908-7924. https://doi.org/10.1109/TVT.2024.3360076
  9. Jacobs, R. A., Jordan, M. I., Nowlan, S. J., & Hinton, G. E. (1991). Adaptive mixtures of local experts. Neural Computation, 3(1), 79-87. https://doi.org/10.1162/neco.1991.3.1.79
  10. Jiralerspong, M., Sun, B., Vucetic, D., Zhang, T., Bengio, Y., Gidel, G., & Malkin, N. (2024). Expected flow networks in stochastic environments and two-player zero-sum games. International Conference on Learning Representations. https://proceedings.iclr.cc/paper_files/paper/2024/hash/2e3eccb54649186564ad6627ed80848c-Abstract-Conference.html
  11. Kanhere, A. V., Gupta, S., Shetty, A., & Gao, G. (2022). Improving GNSS positioning using neural-network-based corrections. NAVIGATION: Journal of the Institute of Navigation, 69(4), Article navi.548. https://doi.org/10.33012/navi.548
  12. Khudoyberdiev, A., Kim, H. Y., & Ryoo, J. (2025). PLUS-CODE+: Zero-installment rover indoor localization. IEEE Sensors Journal, 25(12), 23088-23104. https://doi.org/10.1109/jsen.2025.3561331
  13. Kim, H., Park, J., Park, S., & Ryoo, J. (2022). uGPS: Design and field-tested seamless GNSS infrastructure in metro city. Proceedings of the 28th Annual International Conference on Mobile Computing and Networking, 636-647. https://doi.org/10.1145/3495243.3560520
  14. LabSat. (n.d.). LabSat 4. https://www.labsat.co.uk/en/products/labsat-4
  15. Lange, K. L., Little, R. J. A., & Taylor, J. M. G. (1989). Robust statistical modeling using the t distribution. Journal of the American Statistical Association, 84(408), 881-896. https://doi.org/10.1080/01621459.1989.10478852
  16. Li, J., Xiu, C., Yu, A., Hong, Z., Wang, F., Wang, H., Wang, Y., Chakwizira, J., & Yang, D. (2025). GNSSFormer: Enhancing GNSS single point positioning performance based on transformer for smartphone. IEEE Internet of Things Journal, 12(24), 54304-54319. https://doi.org/10.1109/JIOT.2025.3620847
  17. Li, P., Li, Z., Hou, K., Tang, J., & Xie, S. (2024). Improving GNSS positioning using deep reinforcement learning with selfsupervised learning based data augmentation. Proceedings of the 37th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2024), 2645-2658. https://doi.org/10.33012/2024.19907
  18. Li, Z., Tao, J., Lei, Z., Guo, J., Zhao, Q., & Guo, X. (2025). Factor graph optimization-based RTK/INS integration with raw observations for robust positioning in urban canyons. IEEE Transactions on Instrumentation and Measurement, 74, 1-11. https://doi.org/10.1109/TIM.2025.3577823
  19. Lim, H., Khudoyberdiev, A., Ryu, J., Lee, W. W., Lee, H., & Ryoo, J. (2024). The key to providing positioning information in underground highways for seamless autonomous driving. Korean Society of Civil Engineers Conference, 7-8.
  20. Lou, J., & Zhang, R. (2025). LF-GNSS: Toward more robust satellite positioning with a hard example mining enhanced learning-filtering deep fusion framework. IEEE Transactions on Instrumentation and Measurement, 74, 1-11. https://doi.org/10.1109/TIM.2025.3612654
  21. Luo, S., Li, Y., Liu, S., Zhang, X., Shao, Y., & Wu, C. (2024). Multi-agent continuous control with generative flow networks. Neural Networks, 174, Article 106243. https://doi.org/10.1016/j.neunet.2024.106243
  22. Mnih, V., Kavukcuoglu, K., Silver, D., Rusu, A. A., Veness, J., Bellemare, M. G., Graves, A., Riedmiller, M., Fidjeland, A. K., Ostrovski, G., Petersen, S., Beattie, C., Sadik, A., Antonoglou, I., King, H., Kumaran, D., Wierstra, D., Legg, S., & Hassabis, D. (2015). Human-level control through deep reinforcement learning. Nature, 518(7540), 529-533. https://doi.org/10.1038/nature14236
  23. Mohanty, A., & Gao, G. (2023). Learning GNSS positioning corrections for smartphones using graph convolution neural networks. NAVIGATION: Journal of the Institute of Navigation, 70(4), Article navi.622. https://doi.org/10.33012/navi.622
  24. Nordic Semiconductor. (n.d.). nRF9160 DK. https://www.nordicsemi.com/Products/Development-hardware/nRF9160-DK NovAtel. (n.d.-a). ISA-100C. https://novatel.com/products/gnss-inertial-navigation-systems/imus/isa-100c NovAtel. (n.d.-b). PwrPak7. https://novatel.com/products/receivers/enclosures/pwrpak7
  25. Paek, D.-H., & Kong, S.-H. (2025). Availability-aware sensor fusion via unified canonical space. In D. Belgrave, C. Zhang, H. Lin, R. Pascanu, P. Koniusz, M. Ghassemi, & N. Chen (Eds.), Advances in neural information processing systems (pp. 89259-89277, Vol. 38). Curran Associates, Inc. https://doi.org/10.52202/085713-2986
  26. Pan, L., Zhang, D., Jain, M., Huang, L., & Bengio, Y. (2023). Stochastic generative flow networks. In R. J. Evans & I. Shpitser (Eds.), Proceedings of the thirty-ninth conference on uncertainty in artificial intelligence (pp. 1628-1638, Vol. 216). PMLR. https://proceedings.mlr.press/v216/pan23a.html
  27. Park, K., Kim, Y., Kim, D., & Choi, J. W. (2025). Resilient sensor fusion under adverse sensor failures via multi-modal expert fusion. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 6720-6729. https://doi.org/10.1109/CVPR52734.2025.00630
  28. Park, S., Khudoyberdiev, A., Kim, I., Lee, W. W., Lee, H., & Ryoo, J. (2024). Prevention of tunnel mode entry through resolution of uncertainties in GPS shadow areas for continuous LTE-V2X service. KSCE 2024 Convention, 5-6. https://www.dbpia.co.kr/journal/articleDetail?nodeId=NODE12088857
  29. Perez, E., Strub, F., de Vries, H., Dumoulin, V., & Courville, A. (2018). FiLM: Visual reasoning with a general conditioning layer. Proceedings of the AAAI Conference on Artificial Intelligence, 32(1). https://doi.org/10.1609/aaai.v32i1.11671
  30. Schulman, J., Wolski, F., Dhariwal, P., Radford, A., & Klimov, O. (2017). Proximal policy optimization algorithms. arXiv. https://doi.org/10.48550/arXiv.1707.06347
  31. Stoler, B., Yang, J., Francis, J., & Oh, J. (2025). RCG: Safety-critical scenario generation for robust autonomous driving via real-world crash grounding. arXiv. https://doi.org/10.48550/arXiv.2507.10749
  32. Tang, J., Li, Z., Yu, Q., Zhao, H., Zeng, K., Zhong, S., Wang, Q., Xie, K., Kuzin, V., & Xie, S. (2025). Deep reinforcement learning with robust augmented reward sequence prediction for improving GNSS positioning. GPS Solutions, 29(2), Article 65. https://doi.org/10.1007/s10291-025-01824-wu-blox.(n.d.).EVK-F9DR.https://www.u-blox.com/en/product/evk-f9dr
  33. Wang, J., Xu, B., Liu, J., Gao, K., & Zhang, S. (2024). Sky-GVIO: Enhanced GNSS/INS/vision navigation with FCN-based sky segmentation in urban canyon. Remote Sensing, 16(20), Article 3785. https://doi.org/10.3390/rs16203785
  34. Wang, Y., Huang, H., Xu, Z., Sun, K., Guo, B., & Zhao, J. (2025). Learning from risk: LLM-guided generation of safety-critical scenarios with prior knowledge. arXiv. https://doi.org/10.48550/arXiv.2511.20726
  35. Weng, X., Ling, K. V., & Liu, H. (2024). PrNet: A neural network for correcting pseudoranges to improve positioning with Android raw GNSS measurements. IEEE Internet of Things Journal, 11(14), 24973-24983. https://doi.org/10.1109/JIOT.2024.3392302
  36. Weng, X., Ling, K. V., Liu, H., & Cao, K. (2024). Towards end-to-end GPS localization with neural pseudorange correction. 2024 27th International Conference on Information Fusion (FUSION), 1-7. https://doi.org/10.23919/FUSION59988.2024.10706359
  37. Weng, X., Ling, K. V., Liu, H., Wang, B., & Cao, K. (2026). NeRC: Neural ranging correction through differentiable moving horizon location estimation. Proceedings of the 2026 ACM/IEEE International Conference on Embedded Artificial Intelligence and Sensing Systems, 903-917. https://doi.org/10.1145/3774906.3800475
  38. Whitammer, E. S., Jain, M., Bengio, E., Sun, C., & Bengio, Y. (2022). Trajectory balance: Improved credit assignment in GFlowNets. In S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, & A. Oh (Eds.), Advances in neural information processing systems (pp. 5955-5967, Vol. 35). Curran Associates, Inc. https://doi.org/10.52202/068431-0431
  39. Won, H.-M., Lee, J., & Oh, J. (2025). Localization meets uncertainty: Uncertainty-aware multi-modal localization. Technologies, 13(9), Article 386. https://doi.org/10.3390/technologies13090386
  40. Xia, X., Wen, W., & Hsu, L.-T. (2024). Integrity-constrained factor graph optimization for GNSS positioning in urban canyons. NAVIGATION: Journal of the Institute of Navigation, 71(3), Article navi.660. https://doi.org/10.33012/navi.660
  41. Zeng, K., Li, Z., Zhao, H., Xie, K., Xie, S., Niyato, D., Chen, W., & Zheng, Z. (2024). A spatiotemporal information-driven crossattention model with sparse representation for GNSS NLOS signal classification. IEEE Internet of Things Journal, 11(19), 31892-31908. https://doi.org/10.1109/JIOT.2024.3423016
  42. Zhang, H., Wang, Z., & Vallery, H. (2023). Learning-based NLOS detection and uncertainty prediction of GNSS observations with transformer-enhanced LSTM network. 2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC), 910-917. https://doi.org/10.1109/ITSC57777.2023.10422672
  43. Zhao, H., Li, Z., Wang, Q., Xie, K., Xie, S., Liu, M., & Chen, C. (2024). Improving performances of GNSS positioning correction using multiview deep reinforcement learning with sparse representation. GPS Solutions, 28(3), Article 98. https://doi.org/10.1007/s10291-024-01626-6
  44. Zheng, Z., Wang, Y., & Zhang, Y. (2024). Generating diverse dangerous scenarios considering real-time adversarial reward for autonomous driving test using deep reinforcement learning. 2024 8th CAA International Conference on Vehicular Control and Intelligence (CVCI), 1-6. https://doi.org/10.1109/CVCI63518.2024.10830104