MambaFusion: Adaptive State-Space Fusion for Multimodal 3D Object Detection

TL;DR

MambaFusion is a novel adaptive multi-modal 3D object detection framework that combines efficient state-space models, dynamic feature fusion, and structure-conditioned diffusion to achieve state-of-the-art performance in autonomous driving.

Contribution

It introduces a unified framework integrating selective state-space models, multi-modal token alignment, and diffusion-based reasoning for improved 3D perception.

Findings

Achieves new state-of-the-art on nuScenes benchmarks.

Operates with linear-time complexity.

Provides robust and interpretable 3D detection results.

Abstract

Reliable 3D object detection is fundamental to autonomous driving, and multimodal fusion algorithms using cameras and LiDAR remain a persistent challenge. Cameras provide dense visual cues but ill posed depth; LiDAR provides a precise 3D structure but sparse coverage. Existing BEV-based fusion frameworks have made good progress, but they have difficulties including inefficient context modeling, spatially invariant fusion, and reasoning under uncertainty. We introduce MambaFusion, a unified multi-modal detection framework that achieves efficient, adaptive, and physically grounded 3D perception. MambaFusion interleaves selective state-space models (SSMs) with windowed transformers to propagate the global context in linear time while preserving local geometric fidelity. A multi-modal token alignment (MTA) module and reliability-aware fusion gates dynamically re-weight camera-LiDAR features based on spatial confidence and calibration consistency. Finally, a structure-conditioned diffusion head integrates graph-based reasoning with uncertainty-aware denoising, enforcing physical plausibility, and calibrated confidence. MambaFusion establishes new state-of-the-art performance on nuScenes benchmarks while operating with linear-time complexity. The framework demonstrates that coupling SSM-based efficiency with reliability-driven fusion yields robust, temporally stable, and interpretable 3D perception for real-world autonomous driving systems.