RayMamba: Ray-Aligned Serialization for Long-Range 3D Object Detection

TL;DR

RayMamba introduces a geometry-aware serialization method for voxel-based 3D detectors, enhancing long-range detection by preserving contextual neighborhoods in sparse LiDAR data, with consistent improvements shown in experiments.

Contribution

It proposes a novel ray-aligned serialization strategy that improves long-range 3D detection performance in sparse scenes, compatible with various detectors.

Findings

Achieves up to 2.49 mAP and 1.59 NDS gain on nuScenes at 40-50m range.

Improves VoxelNeXt performance on Argoverse 2 from 30.3 to 31.2 mAP.

Demonstrates consistent improvements across multiple benchmarks.

Abstract

Long-range 3D object detection remains challenging because LiDAR observations become highly sparse and fragmented in the far field, making reliable context modeling difficult for existing detectors. To address this issue, recent state space model (SSM)-based methods have improved long-range modeling efficiency. However, their effectiveness is still limited by generic serialization strategies that fail to preserve meaningful contextual neighborhoods in sparse scenes. To address this issue, we propose RayMamba, a geometry-aware plug-and-play enhancement for voxel-based 3D detectors. RayMamba organizes sparse voxels into sector-wise ordered sequences through a ray-aligned serialization strategy, which preserves directional continuity and occlusion-related context for subsequent Mamba-based modeling. It is compatible with both LiDAR-only and multimodal detectors, while introducing only modest overhead. Extensive experiments on nuScenes and Argoverse 2 demonstrate consistent improvements across strong baselines. In particular, RayMamba achieves up to 2.49 mAP and 1.59 NDS gain in the challenging 40--50 m range on nuScenes, and further improves VoxelNeXt on Argoverse 2 from 30.3 to 31.2 mAP.