ICP-LSTM Joint Localisation for Autonomous Driving
Promet - Traffic&Transportation · 29.06.2026 · 38 (6) · 1357-1373
DOI: 10.7307/ptt.v38i6.1232 ↗Kurz gefasst
Fachartikel in der Zeitschrift Promet - Traffic&Transportation, online erschienen am 29.06.2026. Verfasst von Junhe LIU. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
Autonomous driving requires precise vehicle localisation, especially in moving, unstructured environments. Existing methods focus on spatial matching or temporal sequence modelling for sensor noise, occlusions and partial data. This causes errors. Spatial-only methods struggle with environmental uncertainty, while temporal-only models lack geometric coherence. This study proposes the iterative closest point and long short-term memory-based joint localisation algorithm (ICP-LSTM-JLA) to solve these limitations. This hybrid approach combines geometric point cloud registration and learning-based motion sequence modelling. LSTM-driven predictions that employ inertial and motion data modify the system’s initial posture estimation. Aligning real-time sensor data with pre-mapped environments does this. ICP-LSTM-JLA can enhance localisation accuracy and resilience in diverse driving conditions by integrating spatial alignment with temporal dynamics. The experiments showed that compared to state-of-the-art models, the absolute trajectory error (ATE) has decreased by 17.8%, the relative positioning error (RPE) by 14.3%, the map registration error by 19.1% and the orientation error by 12.6%. This study found that the hybrid architecture improves short-term stability and long-term trajectory accuracy. Thus, it is a reliable real-world autonomous navigation system.
Zitieren
APA
LIU, J. (2026). ICP-LSTM Joint Localisation for Autonomous Driving. Promet - Traffic&Transportation, 38(6), 1357-1373. https://doi.org/10.7307/ptt.v38i6.1232
Vancouver
LIU J. ICP-LSTM Joint Localisation for Autonomous Driving. Promet - Traffic&Transportation. 2026;38(6):1357-1373. doi:10.7307/ptt.v38i6.1232
BibTeX
@article{LIU2026,
author = {LIU, Junhe},
title = {ICP-LSTM Joint Localisation for Autonomous Driving},
journal = {Promet - Traffic&Transportation},
year = {2026},
volume = {38},
number = {6},
pages = {1357--1373},
doi = {10.7307/ptt.v38i6.1232}
}
EUVIMED