This paper develops the first Vision-Language Model (VLM)-based framework for GNSS spoofing detection for autonomous vehicles by fusing front-camera visual data with in-vehicle sensor readings against GNSS-derived maneuvers.
Abstract
Autonomous vehicles (AVs) depend on Global Navigation Satellite Systems (GNSS) for localization and navigation, making them vulnerable to spoofing attacks that can covertly redirect vehicles or induce unsafe maneuvers. In this paper, we develop the first Vision-Language Model (VLM)-based framework for GNSS spoofing detection for autonomous vehicles by fusing front-camera visual data with in-vehicle sensor readings (e.g., speed, acceleration, yaw rate) against GNSS-derived maneuvers. Our approach introduces a three-stage fine-tuning process that first grounds visual cues, and then calibrates sensor data within a shared semantic space to detect discrepancies between predicted and GNSS-derived maneuvers across three attack scenarios. We also generated an independent real-world dataset by driving an instrumented vehicle on public roads in Tuscaloosa, Alabama, equipped with time-synchronized GNSS, IMU, and camera logs to validate cross-regional generalization of our fine-tuned model on unseen data from training data. On this dataset, we then generated intelligent spoofing attacks, including trajectory mirroring with road-network snapping for wrong-turn attacks, position freezing for overshoot scenarios, and drift generation for stop attacks. On this validation dataset, the zero-shot VLMs baseline F1-score ranges from 23% to 32%, whereas our fine-tuned model achieves an F1-score ranging from 94% to 95%. Results show that our VLM-based approach correctly classified every wrong-turn and stop attacks, and attains 88%-93% accuracy for overshoot attacks. Furthermore, we introduce an adaptive inference policy that reduces VLM invocations to 14% (~86% computational reduction) and yields 65ms-73ms per 4s window. These results point to a practical, on-road layer of defense that complements signal-level integrity checks with the use of VLMs.
The presented framework can detect and classify GNSS spoofing attacks in real-time while requiring relatively low computational and memory resources, and is therefore suitable for deployment on resource-constrained vehicular computing platforms.
Abyad Enan, Sagar Dasgupta, Mizanur Rahman et al.· 0 citations
A causal high-order liquid evidence framework for GNSS spoofing detection that achieves the highest F1-scores among the evaluated temporal models on Dataset~1 and Dataset~3, and detects both labeled normal-to-attack transitions within four sampling steps.
Muhammad Ayub Sabir, Junbiao Pang, Fatima Ashraf· 0 citations
Evaluated on GTSRB and LISA across four backbones and three physical attack types, LAMDA is the only method among ten evaluated that consistently improves robustness across all attack-backbone-dataset combinations, while preserving or improving clean accuracy in nearly all cases.
Pedram MohajerAnsari, Amir Salarpour, Mert D. Pesé· 0 citations
Vision-Language Navigation (VLN) enables UAV autonomous navigation in unknown environments by mapping language instructions to real-time visual inputs. Compared with GPS-dependent or pre-programmed navigation, VLN supports intuitive human-machine interaction and stronger environmental adaptability, requiring tight integration of high-level semantic reasoning and low-latency flight control.Existing methods suffer from structural misalignment between global multimodal understanding and sequential action generation, causing jittery trajectories and severe decision latency for long-horizon aerial navigation. To solve this issue, we propose FSD-VLN, a fast-slow dual-system architecture disentangling semantic reasoning and low-latency flight command generation.The framework has two asynchronous branches: a slow stream extracting stable semantic priors from pre-trained vision-language models, and a Diffusion Transformer (DiT) fast stream modeling cross-temporal action distributions to produce consistent flight outputs. We further introduce a time-aware adaptive optimizer to stabilize long-sequence training and reduce gradient oscillation.Large-scale low-altitude simulation experiments show FSD-VLN achieves up to 2X higher navigation success rates on unseen scenes than SOTA methods, while cutting single-action inference delay and total task runtime by over 50%. Our work validates the benefit of decoupled semantic-control modeling and provides a practical paradigm for long-horizon aerial VLN.
Xueke Zhu, Qingyan Meng, Liutao Yu et al.· 1 citation
Objectives: To develop a reliable multi-vehicle detection-based tracking framework under unregulated traffic scenario using enhanced vision transformer model that can precisely localize and track the vehicles. Method: The proposed framework involves a three-level enhancements, (1) Adaptive frame slicing technique in the input data loader to capture the discriminative features of near and far away vehicles in the video frame (2) Geometric positional encoding to provide clear spatial cues and fusion of features, allowing the model to better separate and detect multi-scale vehicles (3) Dual inferencing of sliced and full frame with DIoU-NMS post-processing technique to remove redundant detections. The proposed framework is evaluated on our custom developed Active Learning based vehicle dataset, “AU-INV-P-PALS”, and two public traffic datasets- “IITM-HeTra and Udacity-Fixed small”. Findings: In comparison with the existing baseline DETR model, the proposed model with SGD optimization showed 9.2% increase in the average precision for IoU = 0.5:0.95 indicating its precise localization ability. The model localizes vehicles that occupy as little as 0.2% of the surveillance video frame area. The observed tracking accuracy on real-time traffic CCTV videos indicates that the model can adapt to unregulated traffic environment and achieved appreciable score of about 85.4%, maintaining the tracking stability. The average F1-score of the enhanced DETR obtained for the test frames indicate the generalization ability of the model across different frames. Novelty: The proposed work introduces adaptive frame slicing scheme in the input data loader and geometric positional encoding which enables the detection of faraway vehicles with high accuracy in wide area surveillance imagery. Further, the enhanced model is trained using active learning based custom developed vehicle dataset that significantly reduces the reliance on massive datasets.
Keywords: Deep Learning, Active Learning, Unregulated Road Traffic Control, Vision Transformer
M. Ilamathi, Sabitha Ramakrishnan· Indian Journal of Science an...· 0 citations
Deploying vision-language models (VLMs) for safety-critical spatial reasoning on resource-constrained autonomous driving platforms requires both compact model size and reliable metric grounding. We present MoRAL (Multimodal Reasoning for Autonomous Language Models), a two-stage fine-tuning pipeline that teaches Cosmos-Reason2-2B to first read a physics-encoded Bird's Eye View (BEV) representation and then reason over it for driving decisions. The BEV image encodes LiDAR metric distance as color bands, object class as cluster morphology, and radar Doppler velocity as directional wedge overlays, externalizing spatial perception into the input image so that no learned 3D backbone is required at inference. Stage 1 fine-tunes the vision encoder on 60,000 grounding records; zero-shot baselines produce no parseable BEV outputs, confirming the vocabulary requires explicit training. Stage 2 fine-tunes the full model (52M parameters, 2.4% of total) on 57,696 chain-of-thought records generated by Cosmos-Reason2-8B as teacher, spanning eight driving question types. On 2,304 held-out nuScenes frames evaluated by Gemma 4 (31B) calibrated against human review, MoRAL wins seven of eight question types over a zero-shot 8B baseline despite using four times fewer parameters, with the largest margins on question types requiring structured multi-step physics reasoning. Emergency braking recall improves from 10.8% to 47.8%, output degeneration falls from 94.1% to 20.8%, and the full pipeline fits a consumer 8 GB GPU at 42 tok/s without quantization. These results establish a reproducible foundation for compact, physics-grounded VLM reasoning on mobile edge platforms.
Ambarish Govindarajulu Kaliamurthi, Kai Liu· 0 citations
