UniFlow: Towards Zero-Shot LiDAR Scene Flow for Autonomous Vehicles via Cross-Domain Generalization

2025-11-22

AI Summary

This work addresses the zero-shot cross-sensor generalization challenge in LiDAR-based scene flow estimation—specifically, the inability of existing methods to generalize to unseen LiDAR sensors without target-domain annotations. To overcome domain shifts arising from sensor-specific point cloud distributions, we propose UniFlow, a family of feed-forward networks that learns universal 3D motion priors through joint training on large-scale, multi-source datasets (e.g., Waymo, nuScenes). Crucially, UniFlow explicitly models domain-invariant motion patterns, enabling zero-shot transfer to previously unseen LiDAR configurations—demonstrated for the first time on TruckScenes. Experiments show consistent improvements: +5.1% and +35.2% accuracy on known domains (Waymo and nuScenes, respectively), and a +30.1% gain over dedicated supervised baselines on the entirely unseen TruckScenes—establishing new state-of-the-art performance. This advancement significantly enhances the robustness and adaptability of autonomous driving systems under unknown sensor deployments.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Motion & Tracking

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

LiDAR scene flow is the task of estimating per-point 3D motion between consecutive point clouds. Recent methods achieve centimeter-level accuracy on popular autonomous vehicle (AV) datasets, but are typically only trained and evaluated on a single sensor. In this paper, we aim to learn general motion priors that transfer to diverse and unseen LiDAR sensors. However, prior work in LiDAR semantic segmentation and 3D object detection demonstrate that naively training on multiple datasets yields worse performance than single dataset models. Interestingly, we find that this conventional wisdom does not hold for motion estimation, and that state-of-the-art scene flow methods greatly benefit from cross-dataset training. We posit that low-level tasks such as motion estimation may be less sensitive to sensor configuration; indeed, our analysis shows that models trained on fast-moving objects (e.g., from highway datasets) perform well on fast-moving objects, even across different datasets. Informed by our analysis, we propose UniFlow, a family of feedforward models that unifies and trains on multiple large-scale LiDAR scene flow datasets with diverse sensor placements and point cloud densities. Our frustratingly simple solution establishes a new state-of-the-art on Waymo and nuScenes, improving over prior work by 5.1% and 35.2% respectively. Moreover, UniFlow achieves state-of-the-art accuracy on unseen datasets like TruckScenes, outperforming prior TruckScenes-specific models by 30.1%.

Problem

• Estimating 3D motion between consecutive LiDAR point clouds
• Achieving cross-domain generalization for diverse unseen LiDAR sensors
• Unifying multiple datasets to improve scene flow accuracy

Innovation

• Trains cross-dataset for general motion priors
• Unifies multiple LiDAR datasets with diverse sensors
• Achieves state-of-the-art accuracy on unseen datasets