MR3D-Net: Dynamic Multi-Resolution 3D Sparse Voxel Grid Fusion for LiDAR-Based Collective Perception
This addresses bandwidth and performance issues in collective perception for automated vehicles, representing an incremental improvement over existing fusion methods.
The paper tackles the problem of fusing exchanged information in LiDAR-based collective perception for automated vehicles, proposing MR3D-Net, which achieves state-of-the-art performance on the OPV2V 3D object detection benchmark while reducing required bandwidth by up to 94% compared to early fusion.
The safe operation of automated vehicles depends on their ability to perceive the environment comprehensively. However, occlusion, sensor range, and environmental factors limit their perception capabilities. To overcome these limitations, collective perception enables vehicles to exchange information. However, fusing this exchanged information is a challenging task. Early fusion approaches require large amounts of bandwidth, while intermediate fusion approaches face interchangeability issues. Late fusion of shared detections is currently the only feasible approach. However, it often results in inferior performance due to information loss. To address this issue, we propose MR3D-Net, a dynamic multi-resolution 3D sparse voxel grid fusion backbone architecture for LiDAR-based collective perception. We show that sparse voxel grids at varying resolutions provide a meaningful and compact environment representation that can adapt to the communication bandwidth. MR3D-Net achieves state-of-the-art performance on the OPV2V 3D object detection benchmark while reducing the required bandwidth by up to 94% compared to early fusion. Code is available at https://github.com/ekut-es/MR3D-Net