IJLTEMAS International Journal of Latest Technology in Engineering, Management & Applied Science (IJLTEMAS) IJLTEMAS 2278-2540 IJLTEMAS 618 10.51583/IJLTEMAS.2026.151000097 Artificial Intelligence Adaptive-D3CTTA: Domain-Aware Prototype Learning and Temporal-Geometric Consistency for Continual Test-Time Adaptation of 3D LiDAR Semantic Segmentation Rahavendra Palthiya Department of Computer Science and Engineering Sree Vahini Institute of Science and Technology Tiruvuru, Andhra Pradesh, India India Y. Venkata Raghavarao Dr. Department of Computer Science and Engineering Sree Vahini Institute of Science and Technology Tiruvuru, Andhra Pradesh, India India Veeraswami Pilli Department of Computer Science and Engineering Sree Vahini Institute of Science and Technology Tiruvuru, Andhra Pradesh, India India Prasad Tadi Department of Computer Science and Engineering Sree Vahini Institute of Science and Technology Tiruvuru, Andhra Pradesh, India India 15 10 1195 1208 14 10 2026 20 10 2026 10 11 2026 3D LiDAR Semantic Segmentation Continual Test-Time Adaptation Domain Shift Prototype Learning Temporal Consistency Geometric Consistency Pseudo-Label Refinement Autonomous Driving Abstract

3D LiDAR semantic segmentation is a fundamental task in autonomous driving, but its performance can deteriorate when deployment conditions differ from those observed during training. Changes in point-cloud density, sensor characteristics, scene geometry, and object distributions create domain shifts that challenge models operating in continuously changing environments. This paper proposes Adaptive-D3CTTA, a domain-aware continual test-time adaptation framework for 3D LiDAR semantic segmentation. The framework combines range- and density-aware prototype learning, domain-conditioned feature decorrelation, temporal memory, K-nearest-neighbor (KNN) geometric consistency, and confidence-aware pseudo-label refinement. Range and density information are used to characterize incoming scans and support class-level feature representation, while temporal memory and local geometric agreement help assess the reliability of predictions across sequential observations. A confidence-based refinement mechanism selects reliable pseudo-labels for online adaptation. The framework is designed to process unlabeled target-domain streams without requiring target annotations during adaptation. Its performance should be evaluated using a clearly defined source-to-target protocol on SemanticKITTI and nuScenes-LiDARSeg, with comparisons against the source-only model and relevant test-time adaptation methods, including D³CTTA. Mean intersection-over-union, per-class intersection-over-union, mean class accuracy, adaptation stability, runtime, and memory consumption provide complementary measures of segmentation quality and deployment cost. The proposed framework aims to improve the reliability of continual adaptation by combining semantic, temporal, and geometric information within a unified pipeline. Its effectiveness must be established through reproducible experiments and component-wise ablation studies

 References J. Behley et al., “SemanticKITTI: A Dataset for Semantic Scene Understanding of LiDAR Sequences,” Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pp. 9297–9307, 2019. H. Caesar et al., “nuScenes: A Multimodal Dataset for Autonomous Driving,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 11621–11631, 2020. J. Zhao, H. Jiang, H. Song, J. Xiao, and D. Gong, “D³CTTA: Domain-Dependent Decorrelation for Continual Test-Time Adaption of 3D LiDAR Segmentation,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 11864–11874, 2025. D. Wang, E. Shelhamer, S. Liu, B. Olshausen, and T. Darrell, “Tent: Fully Test-Time Adaptation by Entropy Minimization,” International Conference on Learning Representations (ICLR), 2021. Q. Wang, O. Fink, L. Van Gool, and D. Dai, “Continual Test-Time Domain Adaptation,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 7201–7211, 2022. S. Niu et al., “Efficient Test-Time Model Adaptation without Forgetting,” Proceedings of the International Conference on Machine Learning (ICML), vol. 162, pp. 16888–16905, 2022. C. Saltori et al., “GIPSO: Geometrically Informed Propagation for Online Adaptation in 3D LiDAR Segmentation,” European Conference on Computer Vision (ECCV), 2022. Q. Hu et al., “RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 11108–11117, 2020. H. Thomas et al., “KPConv: Flexible and Deformable Convolution for Point Clouds,” Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pp. 6411–6420, 2019. Y. Zhang et al., “PolarNet: An Improved Grid Representation for Online LiDAR Point Clouds Semantic Segmentation,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 9601–9610, 2020. X. Zhu et al., “Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR Segmentation,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 9939–9948, 2021. C. Xu et al., “SqueezeSegV3: Spatially-Adaptive Convolution for Efficient Point-Cloud Segmentation,” European Conference on Computer Vision (ECCV), 2020. A. Milioto, I. Vizzo, J. Behley, and C. Stachniss, “RangeNet++: Fast and Accurate LiDAR Semantic Segmentation,” IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2019. T. Cortinhal, G. Tzelepis, and E. E. Aksoy, “SalsaNext: Fast, Uncertainty-Aware Semantic Segmentation of LiDAR Point Clouds,” International Symposium on Visual Computing (ISVC), 2020. C. R. Qi, H. Su, K. Mo, and L. J. Guibas, “PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017. C. R. Qi, L. Yi, H. Su, and L. J. Guibas, “PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space,” Advances in Neural Information Processing Systems (NeurIPS), 2017. Y. Wang et al., “Dynamic Graph CNN for Learning on Point Clouds,” ACM Transactions on Graphics, vol. 38, no. 5, 2019. B. Graham, M. Engelcke, and L. van der Maaten, “3D Semantic Segmentation with Submanifold Sparse Convolutional Networks,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2018. C. Choy, J. Gwak, and S. Savarese, “4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019. X. Wu et al., “SqueezeSegV2: Improved Model Structure and Unsupervised Domain Adaptation for Road-Object Segmentation from a LiDAR Point Cloud,” IEEE International Conference on Robotics and Automation (ICRA), 2019. J. Liang, D. Hu, and J. Feng, “Do We Really Need to Access the Source Data? Source Hypothesis Transfer for Unsupervised Domain Adaptation,” Proceedings of the International Conference on Machine Learning (ICML), vol. 119, pp. 6028–6039, 2020. Y. Ganin et al., “Domain-Adversarial Training of Neural Networks,” Journal of Machine Learning Research, vol. 17, no. 59, pp. 1–35, 2016. B. Sun and K. Saenko, “Deep CORAL: Correlation Alignment for Deep Domain Adaptation,” European Conference on Computer Vision (ECCV) Workshops, 2016. A. Tarvainen and H. Valpola, “Mean Teachers Are Better Role Models: Weight-Averaged Consistency Targets Improve Semi-Supervised Deep Learning Results,” Advances in Neural Information Processing Systems (NeurIPS), 2017. K. He et al., “Momentum Contrast for Unsupervised Visual Representation Learning,” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020.