Awesome

MD-LVIO

This repo belongs to my own postgraduate research project, so the whole project will be opensource until I graduate from SEU !!

Dynamic-Aware LIO with GPU Accelaration (Half-done)
Graph-Based Relocalization (Done)
Multi-Session Map Fusion (Done)
Object-Level Detection and Update (Todo)
Application (Done)

please see https://github.com/Yixin-F/better_fastlio2

0. Our sensing platform

1. Dynamic-Aware LIO with GPU Accelaration

The main contributions are as follows:

We propose a novel egocentric descriptor called Segmented Curved-Voxel Occupancy Descriptor (SCV-OD), based on which a dynamic-aware and LiDAR-only SLAM framework is built for mobile sensing with various sensor types in unknown and complex environments.
We propose a multi-stage object segmentation method assisted with LiDAR intensity and geometric features. This method is tightly coupled with a lightweight object tracking approach, allowing for the effective removal of high dynamic objects and the refinement of imperfect objects through curved-voxel occupancy detection.
We modify VGICP as the pipeline for the SCV-OD, efficiently estimating initial motions at 10 Hz and update LiDAR poses with considering dynamic removal at 15 -20 Hz. Consequently, a high-quality static instance map in the global coordinate system is generated.
The proposed method is quantitatively and qualitatively evaluated on the KITTI dataset and the custom dataset. Finally, we demonstrate that our proposed approach outperforms the state-of-the-art studies in static mapping.

1) System Overview

2) Evaluation on Kitti

3) Evaluation on Our Custom Dataset

2. Graph-Based Relocalization

Unlike most of the existing methods that focus on extracting local, global, and statistical features of raw point clouds, our method aims at the semantic level that can be superior in terms of robustness to environmental changes. Inspired by the perspective of humans, who recognize scenes through identifying semantic objects and capturing their relations, this paper presents a novel semantic graph based approach for place recognition. First, we propose a novel semantic graph representation for the point cloud scenes by reserving the semantic and topological information of the raw point cloud. Thus, place recognition is modeled as a graph matching problem. Then we design a fast and effective graph similarity network to compute the similarity. Exhaustive evaluations on the KITTI dataset show that our approach is robust to the occlusion as well as viewpoint changes and outperforms the state-of-theart methods with a large margin.

1) Framework

2) Evaluation on Kitti

3. Multi-Session Map Fusion

Long-term 3D map management is a fundamental capability required by a robot to reliably navigate in the non-stationary real-world. This paper develops open-source, modular, and readily available LiDAR-based lifelong mapping for urban sites. This is achieved by dividing the problem into successive subproblems: multi-session SLAM (MSS), high/low dynamic change detection, and positive/negative change management. The proposed method leverages MSS and minimizes potential trajectory error; thus, a manual or good initial alignment is not required for change detection. Our change management scheme preserves efficacy in both memory and computation costs, providing automatic object segregation from a large-scale point cloud map.

1) SLAM Result in Different Session for Parkinglot

2) Pose Graph Constrain Using Isam2

3) Map Fusion

4. Object-Level Detection and Update

It's still under research, so comming soon !! But you can find a simple version in map_generation and difference_detection

5. Application

Applied to urban power distribution road scanning, just show the result

1) Sence and Equipments

2) Result