GLiDR: Topologically Regularized Graph Generative Network for Sparse LiDAR Point Clouds

TL;DR

GLiDR tackles sparse LiDAR challenges by modeling scans as graphs and enforcing a global static topology via $0$-dim $\,mathcal{PH}$ constraints. The method uses a four-layer LiDAR graph encoder with a 512-dimensional latent space and a topology-aware loss to create a single connected backbone, guiding augmentation along the global backbone. It relies on a range-image representation to enable simple $AE$-based losses, avoiding costly $CD$/$EMD$, and demonstrates superior static-point augmentation and SLAM performance across KITTI, ARD-16, and CARLA, including $32 imes$ sparser dynamics. GLiDR also yields accurate binary masks for static/dynamic segmentation, highlighting practical value for navigation and safety in constrained environments.

Abstract

Sparse LiDAR point clouds cause severe loss of detail of static structures and reduce the density of static points available for navigation. Reduced density can be detrimental to navigation under several scenarios. We observe that despite high sparsity, in most cases, the global topology of LiDAR outlining the static structures can be inferred. We utilize this property to obtain a backbone skeleton of a LiDAR scan in the form of a single connected component that is a proxy to its global topology. We utilize the backbone to augment new points along static structures to overcome sparsity. Newly introduced points could correspond to existing static structures or to static points that were earlier obstructed by dynamic objects. To the best of our knowledge, we are the first to use such a strategy for sparse LiDAR point clouds. Existing solutions close to our approach fail to identify and preserve the global static LiDAR topology and generate sub-optimal points. We propose GLiDR, a Graph Generative network that is topologically regularized using 0-dimensional Persistent Homology ($\mathcal{PH}$) constraints. This enables GLiDR to introduce newer static points along a topologically consistent global static LiDAR backbone. GLiDR generates precise static points using $32\times$ sparser dynamic scans and performs better than the baselines across three datasets. GLiDR generates a valuable byproduct - an accurate binary segmentation mask of static and dynamic objects that are helpful for navigation planning and safety in constrained environments. The newly introduced static points allow GLiDR to outperform LiDAR-based navigation using SLAM in several settings. Source code is available at https://kshitijbhat.github.io/glidr

Figures (6)

• Figure 1: GLiDR outperforms the best baselines on SLAM results with sparse LiDAR scans. Binary segmentation masks of dynamic and static points (blue denotes dynamic and red denotes static points) obtained using GLiDR are also better.
• Figure 2: A sample LiDAR graph layer. The input LiDAR is transformed into a graph using k-nearest neighbors (k=20). Graph convolution extracts neighborhood features followed by max-pool transform the LiDAR embedding to a higher dimensional space.
• Figure 3: LiDAR Graph Generative Model using stacked LiDAR Graph layers
• Figure 4: GLiDR network architecture with the generative LiDAR graph layers (upper branch). The lower branches constitute the 0-dim $\mathcal{PH}$ constraints and generate the topological loss.
• Figure 5: Left: Input Dynamic LiDAR Middle: Visualization of the $0$-dim $\mathcal{PH}$ based LiDAR backbone after some epochs. Right: LiDAR backbone at the end as a single connected component.