LIORNet: Self-Supervised LiDAR Snow Removal Framework for Autonomous Driving under Adverse Weather Conditions
Authors: Ji-il Park, Inwook Shim
Organizations: Ministry of National Defense, 22, Itaewon-ro, Yongsan-gu, Seoul, Republic of Korea. · Department of Smart Mobility Engineering, Inha University, Republic of Korea.
LiDAR sensors provide high-resolution 3D perception and long-range detection, making them indispensable for autonomous driving and robotics. However, their performance significantly degrades under adverse weather conditions such as snow, rain, and fog, where spurious noise points dominate the point cloud and lead to false perception. To address this problem, various approaches have been proposed: distance-based filters exploiting spatial sparsity, intensity-based filters leveraging reflectance distributions, and learning-based methods that adapt to complex environments. Nevertheless, distance-based methods struggle to distinguish valid object points from noise, intensity-based methods often rely on fixed thresholds that lack adaptability to changing conditions, and learning-based methods suffer from the high cost of annotation, limited generalization, and computational overhead. In this study, we propose LIORNet, which eliminates these drawbacks and integrates the strengths of all three paradigms. LIORNet is built upon a U-Net++ backbone and employs a self-supervised learning strategy guided by pseudo-labels generated from multiple physical and statistical cues, including range-dependent intensity thresholds, snow reflectivity, point sparsity, and sensing range constraints. This design enables LIORNet to distinguish noise points from environmental structures without requiring manual annotations, thereby overcoming the difficulty of snow labeling and the limitations of single-principle approaches. Extensive experiments on the WADS and CADC datasets demonstrate that LIORNet outperforms state-of-the-art filtering algorithms in both accuracy and runtime while preserving critical environmental features. These results highlight LIORNet as a practical and robust solution for LiDAR perception in extreme weather, with strong potential for real-time deployment in autonomous driving systems.
Figures & tables
Fig. 1: Categorization by approach [ 8 ] . LIORNet shown at the center is the first attempt to unify all three strategies.
Fig. 2: Overall LIORNet framework: (a) data processing pipeline, (b) network architecture, and backbone configurations using (c) U-Net and (d) U-Net++.
Fig. 3: Visualization of conditional pseudo-label point updates. Candidate snow points are sequentially refined through intensity (Cond1), reflectivity (Cond2), edge (Cond3), and sparsity (Cond4) conditions resulting in the final pseudo-labeled snow points.
ID
UW
PP
DS
Precision
Recall (TPR)
FPR
FNR
IoU
1
X
X
X
0.379 ± 0.107
0.942 ± 0.088
0.265 ± 0.072
0.058 ± 0.088
0.368 ± 0.104
2
X
O
X
0.473 ± 0.136
0.961 ± 0.068
0.175 ± 0.049
0.039 ± 0.068
0.462 ± 0.132
3
X
X
O
0.402 ± 0.106
0.939 ± 0.093
0.240 ± 0.069
0.061 ± 0.093
0.389 ± 0.103
4
X
O
O
0.498 ± 0.132
0.958 ± 0.072
0.157 ± 0.046
0.042 ± 0.072
0.486 ± 0.128
5
20%
X
X
0.378 ± 0.104
0.941 ± 0.090
0.266 ± 0.074
0.059 ± 0.090
0.367 ± 0.101
6
20%
O
X
0.477 ± 0.133
0.960 ± 0.070
0.172 ± 0.048
0.040 ± 0.070
0.465 ± 0.129
TABLE I: Quantitative Evaluation of LIORNet under Different Configurations. Postprocess was applied with the theoretical snow sensing limit of 71.23m. (UW: Uncertainty-based Weighting, PP: Postprocessing, DS: Deep Supervision)
Fig. 4: False-positive snow data before and after applying the rule-based postprocessing (considering LiDAR height and the theoretical snow-sensing distance).
Fig. 5: Comparison of performance on the (A) WADS and (B) CADC datasets. (a) Raw, (b) DROR, (c) LIOR, (d) D-LIOR, (e) LiSnowNet, (f) LIORNet.
Method
Precision
Recall
F1-score
F3-score
F5-score
Runtime
ms
Hz
DROR
0.764 ± 0.021
0.745 ± 0.152
0.754
0.747
0.746
3505
0.3
LIOR
0.827 ± 0.027
0.908 ± 0.138
0.866
0.899
0.905
208.1
4.8
D - LIOR
0.918 ± 0.038
0.875 ± 0.125
0.896
0.879
0.877
115.6
8.7
LiSnowNet
0.951 ± 0.039
0.842 ± 0.167
0.893
0.852
0.846
4.98
201.0
LIORNet (U-Net)
0.379 ± 0.107
0.942 ± 0.088
0.541
0.820
0.891
21.39
46.8
TABLE II: Quantitative evaluation on the WADS dataset. Postprocess was applied using the optimal range of 20m.
Configuration
Precision
Recall
Baseline(Full)
0.672 ± 0.132
0.920 ± 0.114
- Intensity term (Cond1)
0.306 ± 0.094
0.951 ± 0.076
- Reflectivity term (Cond2)
0.401 ± 0.110
0.938 ± 0.092
- Edge term (Cond3)
0.195 ± 0.118
0.679 ± 0.136
- Sparsity term (Cond4)
0.387 ± 0.109
0.941 ± 0.090
- Penalty term (Cond5)
0.256 ± 0.099
0.962 ± 0.061
TABLE III: Ablation study on the contributions of each component. The study is conducted based on ID 4 in Table I .
Fig. 6: Performance evaluation on real-world data (a) Raw, (b) DROR, (c) LIOR, (d) D-LIOR, (e) LiSnowNet, (f) LIORNet.
Light Detection and Ranging (LiDAR) is critical for autonomous perception, yet its performance is known to degrade in adverse weather. While existing literature models snow as atmospheric noise that degrades performance, our granular evaluation on the WADS-3D and CADC datasets reveals a counter-intuitive phenomenon: detection accuracy consistently improves as snow intensity increases. Through ablation study and meta-information analysis, we reconcile this by confirming that while falling snow acts as noise, its negative impact is outweighed by a dominant, concurrent factor: significantly increased object point count. We hypothesize that this object point count boost acts as a geometric enhancement, driven by snow accumulation increasing vehicle surface area and a shift from specular to diffuse reflectance. Our findings demonstrate that evaluating perception systems using broad weather-severity bins without accounting for correlated object-level data can lead to misleading conclusions about detector robustness. We conclude that accurate evaluation requires decoupling atmospheric noise from surface accumulation effects and outline future research to validate these physical models. The WADS-3D dataset is available at: https://digitalcommons.mtu.edu/wads-3d/.
Accurate LiDAR simulation is crucial for autonomous driving, especially under adverse weather conditions. Existing methods struggle to capture the complex interactions between LiDAR signals and atmospheric phenomena, leading to unrealistic representations. This paper presents a physics-informed learning framework (PICWGAN) for generating realistic LiDAR data under adverse weather conditions. By integrating physicsdriven constraints for modeling signal attenuation and geometryconsistent degradations into a physics-informed learning pipeline, the proposed method reduces the sim-to-real gap. Evaluations on real-world datasets (CADC for snow, Boreas for rain) and the VoxelScape dataset show that our approach closely mimics realworld intensity patterns. Quantitative metrics, including MSE, SSIM, KL divergence, and Wasserstein distance, demonstrate statistically consistent intensity distributions. Additionally, models trained on data enhanced by our framework outperform baselines in downstream 3D object detection, achieving performance comparable to models trained on real-world data. These results highlight the effectiveness of the proposed approach in improving the realism of LiDAR data and enabling robust perception under adverse weather conditions.
Vivek Anand, Bharat Lohani, Rakesh Mishra +1
Geoinformatics Group, Department of Civil Engineering, Indian Institute of Technology Kanpur, India · Department of Geodesy and Geomatics Engineering, University of New Brunswick, NB, Canada · Department of Engineering Technology & Industrial Distribution (ETID) of Texas A&M University
Autonomous vehicles face major perception and navigation challenges in adverse weather such as rain, fog, and snow, which degrade the performance of LiDAR, RADAR, and RGB camera sensors. While each sensor type offers unique strengths, such as RADAR robustness in poor visibility and LiDAR precision in clear conditions, they also suffer distinct limitations when exposed to environmental obstructions. This study proposes LRC-WeatherNet, a novel multi-sensor fusion framework that integrates LiDAR, RADAR, and camera data for real-time classification of weather conditions. By employing both early fusion using a unified Bird's Eye View representation and mid-level gated fusion of modality-specific feature maps, our approach adapts to the varying reliability of each sensor under changing weather. Evaluated on the extensive MSU-4S dataset covering nine weather types, LRC-WeatherNet achieves superior classification performance and computational efficiency, significantly outperforming unimodal baselines in adverse conditions. This work is the first to combine all three modalities for robust, real-time weather classification in autonomous driving. We release our trained models and source code in https://github.com/nouralhudaalbashir/LRC-WeatherNet.
Nour Alhuda Albashir, Lars Pernickel, Danial Hamoud +2
Halmstad University, School of Information Technology, Center for Applied Intelligent Systems Research, Halmstad, Sweden