CAFuser: Condition-Aware Multimodal Fusion for Robust Semantic Perception of Driving Scenes
This addresses robust semantic perception for autonomous driving systems, offering a novel approach to improve performance in adverse conditions, though it is incremental in enhancing existing fusion techniques.
The paper tackles the problem of suboptimal sensor fusion in autonomous driving by proposing a condition-aware multimodal fusion method, CAFuser, which dynamically adapts fusion based on environmental conditions, achieving state-of-the-art results such as 59.7 PQ on MUSES and 78.2 mIoU for segmentation.
Leveraging multiple sensors is crucial for robust semantic perception in autonomous driving, as each sensor type has complementary strengths and weaknesses. However, existing sensor fusion methods often treat sensors uniformly across all conditions, leading to suboptimal performance. By contrast, we propose a novel, condition-aware multimodal fusion approach for robust semantic perception of driving scenes. Our method, CAFuser, uses an RGB camera input to classify environmental conditions and generate a Condition Token that guides the fusion of multiple sensor modalities. We further newly introduce modality-specific feature adapters to align diverse sensor inputs into a shared latent space, enabling efficient integration with a single and shared pre-trained backbone. By dynamically adapting sensor fusion based on the actual condition, our model significantly improves robustness and accuracy, especially in adverse-condition scenarios. CAFuser ranks first on the public MUSES benchmarks, achieving 59.7 PQ for multimodal panoptic and 78.2 mIoU for semantic segmentation, and also sets the new state of the art on DeLiVER. The source code is publicly available at: https://github.com/timbroed/CAFuser.