RE-TRIP : Reflectivity Instance Augmented Triangle Descriptor for 3D Place Recognition

TL;DR

This paper proposes a novel descriptor for 3D PR, named RE-TRIP (REflectivity-instance augmented TRIangle descriPtor), which leverages both geometric measurements and reflectivity to enhance robustness in challenging scenarios such as geometric degeneracy, high geometric similarity, and the presence of dynamic objects.

Abstract

While most people associate LiDAR primarily with its ability to measure distances and provide geometric information about the environment (via point clouds), LiDAR also captures additional data, including reflectivity or intensity values. Unfortunately, when LiDAR is applied to Place Recognition (PR) in mobile robotics, most previous works on LiDAR-based PR rely only on geometric measurements, neglecting the additional reflectivity information that LiDAR provides. In this paper, we propose a novel descriptor for 3D PR, named RE-TRIP (REflectivity-instance augmented TRIangle descriPtor). This new descriptor leverages both geometric measurements and reflectivity to enhance robustness in challenging scenarios such as geometric degeneracy, high geometric similarity, and the presence of dynamic objects. To implement RE-TRIP in real-world applications, we further propose (1) a keypoint extraction method, (2) a key instance segmentation method, (3) a RE-TRIP matching method, and (4) a reflectivity-combined loop verification method. Finally, we conduct a series of experiments to demonstrate the effectiveness of RE-TRIP. Applied to public datasets (i.e., HELIPR, FusionPortable) containing diverse scenarios such as long corridors, bridges, large-scale urban areas, and highly dynamic environments -- our experimental results show that the proposed method outperforms existing state-of-the-art methods in terms of Scan Context, Intensity Scan Context, and STD.

Figures (5)

• Figure 1: Overview of the proposed method. (a) Keypoints (red points) extracted from the high-reflectivity objects (yellow box). (b) LiDAR scan when revisiting the same location. (c) LiDAR scan from the previous visit. Despite (b) and (c) are captured from different viewpoints due to opposite driving directions, keypoints (red points) are consistently detected from the high-reflectivity objects (yellow box). (d) By associating these distinctive instances into a triangular shape, RE-TRIP effectively retrieves the same location from the database.
• Figure 2: The overall framework of our proposed method consists of four steps. First, keypoints are extracted based on reflectivity measurement, identifying Absolute Reflectivity Points (ARP) and Relative Reflectivity Points (RRP). Then keypoints are grouped into instances and a small number of significant key instances are selected to maintain efficiency while keeping computational costs low. Next, we generate Reflectivity Instance Augmented Triangle Descriptors (RE-TRIP), which incorporate both geometric and reflectivity information. RE-TRIPs are then retrieved from the hash table and matched based on key instance similarity and side lengths. Finally, candidate frames are verified through reflectivity-combined geometric verification.
• Figure 3: Description of RE-TRIP matching between $\bm{\mathcal{D}}^Q$ and $\bm{\mathcal{D}}^R$. It consists of three steps: (1) side length matching, (2) label matching, and (3) instance size comparison.
• Figure 4: Precision-Recall curves (top two rows) and F1 Score-Recall curves (bottom two rows) for HELIPR and FusionPortable.
• Figure 5: Keypoint ablation results. (a) and (b) compare the performance of ARP, RRP, and the combined approach across all sequences of HELIPR. The combined use of both ARP and RRP consistently yields superior results, demonstrating the complementary nature of both keypoints.