Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SuperiorGAT: Graph Attention Networks for Sparse LiDAR Point Cloud Reconstruction in Autonomous Systems

Khalfalla Awedat, Mohamed Abidalrekab, Gurcan Comert, Mustafa Ayad

Abstract

LiDAR-based perception in autonomous systems is constrained by fixed vertical beam resolution and further compromised by beam dropout resulting from environmental occlusions. This paper introduces SuperiorGAT, a graph attention-based framework designed to reconstruct missing elevation information in sparse LiDAR point clouds. By modeling LiDAR scans as beam-aware graphs and incorporating gated residual fusion with feed-forward refinement, SuperiorGAT enables accurate reconstruction without increasing network depth. To evaluate performance, structured beam dropout is simulated by removing every fourth vertical scanning beam. Extensive experiments across diverse KITTI environments, including Person, Road, Campus, and City sequences, demonstrate that SuperiorGAT consistently achieves lower reconstruction error and improved geometric consistency compared to PointNet-based models and deeper GAT baselines. Qualitative X-Z projections further confirm the model's ability to preserve structural integrity with minimal vertical distortion. These results suggest that architectural refinement offers a computationally efficient method for improving LiDAR resolution without requiring additional sensor hardware.

SuperiorGAT: Graph Attention Networks for Sparse LiDAR Point Cloud Reconstruction in Autonomous Systems

Abstract

LiDAR-based perception in autonomous systems is constrained by fixed vertical beam resolution and further compromised by beam dropout resulting from environmental occlusions. This paper introduces SuperiorGAT, a graph attention-based framework designed to reconstruct missing elevation information in sparse LiDAR point clouds. By modeling LiDAR scans as beam-aware graphs and incorporating gated residual fusion with feed-forward refinement, SuperiorGAT enables accurate reconstruction without increasing network depth. To evaluate performance, structured beam dropout is simulated by removing every fourth vertical scanning beam. Extensive experiments across diverse KITTI environments, including Person, Road, Campus, and City sequences, demonstrate that SuperiorGAT consistently achieves lower reconstruction error and improved geometric consistency compared to PointNet-based models and deeper GAT baselines. Qualitative X-Z projections further confirm the model's ability to preserve structural integrity with minimal vertical distortion. These results suggest that architectural refinement offers a computationally efficient method for improving LiDAR resolution without requiring additional sensor hardware.
Paper Structure (15 sections, 10 equations, 7 figures, 1 table)

This paper contains 15 sections, 10 equations, 7 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Graph-Based Reconstruction Methodology
  4. Experimental Setup and Evaluation
  5. Conclusion

Figures (7)

  • Figure 1: LiDAR point cloud with beam dropout (KITTI Residential dataset). Original 64-beam point cloud (colored by z-coordinate) with 25% beam dropout points in red, simulating missing z-coordinates due to hardware or environmental faults. SuperiorGAT reconstructs these z-values for autonomous driving.
  • Figure 2: Illustration of two-layer message passing in a GNN.
  • Figure 3: GAT architecture comparison. (a) Standard implementation. (b) SuperiorGAT with domain-specific enhancements for LiDAR reconstruction.
  • Figure 4: Beam-indexed graph structure used for LiDAR point reconstruction, illustrating intra- and inter-beam connections. Dropped points are marked to show the reconstruction target.
  • Figure 5: Sensitivity analysis of the neighborhood size $k$ illustrating (a) the relationship between vertical reconstruction error ($RMSE_z$) and inference latency, and (b) the corresponding geometric shape fidelity measured by Chamfer distance.
  • ...and 2 more figures