MULTIMODAL SENSOR FUSION FOR AUTONOMOUS DRIVING USING LARGE-SCALE MACHINE LEARNING MODELS
International Journal of Grid Computing & Applications · 28.03.2026 · 17 (1) · 73-83
DOI: 10.5121/ijgca.2026.17106 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Grid Computing & Applications, online erschienen am 28.03.2026. Verfasst von Sungho Kim und 8 weiteren Autor:innen.
Abstract (Originalsprache)
Autonomous driving systems rely on accurate and reliable environment perception to ensure safe navigation in complex and dynamic traffic scenarios. Multimodal sensor fusion has emerged as a fundamental approach to overcoming the limitations of individual sensing modalities such as cameras, LiDAR, and radar. This research investigates the integration of multimodal sensor fusion with large scale machine learning models, including convolutional neural networks (CNNs), transformer based architectures, and foundation models, to enhance 3D object detection and scene understanding. The study formulates multimodal perception as a probabilistic learning problem and explores early, mid level (Bird’s Eye View), and late fusion strategies, with particular emphasis on attention based cross modal mechanisms. Large scale datasets such as nuScenes and Waymo Open Dataset are utilized to evaluate performance under diverse environmental and operational conditions. Experimental results demonstrate that attention driven fusion architectures outperform single modality baselines and traditional fusion strategies in terms of detection accuracy, robustness under adverse weather and low light conditions, and fault tolerance during partial sensor degradation. Furthermore, the research examines scalability, computational efficiency, uncertainty modelling, and real time deployment constraints. The findings indicate that transformer based BEV fusion frameworks provide a scalable and interpretable pathway toward next generation autonomous perception systems. This work contributes a comprehensive analysis of architectural design choices, fusion mechanisms, and deployment considerations, supporting the development of robust, efficient, and safety aware autonomous driving platforms.
Zitieren
APA
Kim, S., Matar, M., Mueses, E. P., Bhardwaj, P., Kim, S., Kim, T., Kyoom, N. M., Alim, A., & Wongsawat, S. (2026). MULTIMODAL SENSOR FUSION FOR AUTONOMOUS DRIVING USING LARGE-SCALE MACHINE LEARNING MODELS. International Journal of Grid Computing & Applications, 17(1), 73-83. https://doi.org/10.5121/ijgca.2026.17106
Vancouver
Kim S, Matar M, Mueses EP, Bhardwaj P, Kim S, Kim T, et al. MULTIMODAL SENSOR FUSION FOR AUTONOMOUS DRIVING USING LARGE-SCALE MACHINE LEARNING MODELS. International Journal of Grid Computing & Applications. 2026;17(1):73-83. doi:10.5121/ijgca.2026.17106
BibTeX
@article{Kim2026,
author = {Kim, Sungho and Matar, Mahmoud and Mueses, Edgar Plasas and Bhardwaj, Parvati and Kim, Seungwon and Kim, Taehyun and Kyoom, Niaz Mahmood and Alim, Abdul and Wongsawat, Sudawan},
title = {MULTIMODAL SENSOR FUSION FOR AUTONOMOUS DRIVING USING LARGE-SCALE MACHINE LEARNING MODELS},
journal = {International Journal of Grid Computing & Applications},
year = {2026},
volume = {17},
number = {1},
pages = {73--83},
doi = {10.5121/ijgca.2026.17106}
}
EUVIMED