Accurate 3D Semantic Segmentation of Point Clouds for Intelligent Vehicles Based on Multi-view Edge Guidance and Fusion

Semantic segmentation based on LiDAR plays an important role in the environment perception, path planning and decision control of unmanned ground systems. However, in challenging environments with complex background interleaving, the high proportion of drivable areas and significant target elements pose serious challenges to segmentation tasks. Existing studies attempt to address these challenges but still grapple with issues like inter-class semantic conflicts and inaccurate edge segmentation. To address these challenges effectively, a reliable point cloud segmentation network driven by multi-view fusion and multi-edge guidance is specially customized. Specifically, the network’s point cloud multi-view RIV (Range Image View) and BEV (Bird’s Eye View) encoding backbones use lightweight pyramid architectures to extract specific details and semantic features of multi-scale point clouds. Subsequently, a dedicated multi-view edge guidance module focuses on enhancing interclass edge differentiation features from RIV and BEV, respectively. These features are fused with the encoding backbone hierarchy, and the extraction of edge information is supervised by a new hybrid edge loss to maximize the consistency of semantic segmentation and predicted edges. Furthermore, a multi-view and multi-scale feature fusion module based on a multi-head attention mechanism is introduced to enhance the potential complementarity and interaction of the captured features. Through comprehensive experiments and ablation studies on the SemanticKITTI dataset and the RELLIS-3D dataset, recognized in the field of autonomous driving, the results demonstrate that our custom method is competitive in terms of IoU accuracy and real-time indicators compared to typical baseline methods.