Pose optimization for autonomous driving datasets using neural rendering models

arXiv 2025

Quentin Herau1,2, Nathan Piasco1, Moussab Bennehar1, Luis Roldão1, Dzmitry Tsishkou1, Bingbing Liu1, Cyrille Migniot3, Pascal Vasseur4, Cédric Demonceaux2
1Noah's Ark Lab, Huawei Technologies
2ICB UMR CNRS 6303, Université de Bourgogne Europe, France 3ImVIA UR 7535, Université de Bourgogne Europe, France 4MIS UR 4290, Université de Picardie Jules Verne, France
arXiv PO Benchmark NVS Benchmark

Abstract

Autonomous driving systems rely on accurate perception and localization of the ego car to ensure safety and reliability in challenging real-world driving scenarios. Public datasets play a vital role in benchmarking and guiding advancement in research by providing standardized resources for model development and evaluation. However, potential inaccuracies in sensor calibration and vehicle poses within these datasets can lead to erroneous evaluations of downstream tasks, adversely impacting the reliability and performance of the autonomous systems. To address this challenge, we propose a robust optimization method based on Neural Radiance Fields (NeRF) to refine sensor poses and calibration parameters, enhancing the integrity of dataset benchmarks. To validate improvement in accuracy of our optimized poses without ground truth, we present a thorough evaluation process, relying on reprojection metrics, Novel View Synthesis rendering quality, and geometric alignment. We demonstrate that our method achieves significant improvements in sensor pose accuracy. By optimizing these critical parameters, our approach not only improves the utility of existing datasets but also paves the way for more reliable autonomous driving models. To foster continued progress in this field, we make the optimized sensor poses publicly available, providing a valuable resource for the research community.
Teaser Image

Dataset pose improvement: Left, rendering comparison between original poses and optimized poses on Waymo open dataset (from left to right: normal map, depth map and RGB rendering). Right, the changes in metrics between the original poses (in blue) and the poses optimized with MOISST (in red) for each dataset, grouped in 3 categories: NVS, SfM, Geometric.

Qualitative Results

BibTeX

@article{herau2025pose,
  title={Pose Optimization for Autonomous Driving Datasets using Neural Rendering Models},
  author={Herau, Quentin and Piasco, Nathan and Bennehar, Moussab and Roldao, Luis and Tsishkou, Dzmitry and Liu, Bingbing and Migniot, Cyrille and Vasseur, Pascal and Demonceaux, C{\'e}dric},
  journal={arXiv preprint arXiv:2504.15776},
  year={2025}
}
}