How to install M-LOAM on Ubuntu 18.04 with ROS?

Follow the prerequisites in the README: install ROS Melodic, run the provided script for Ceres Solver and Eigen3, then build in a catkin workspace. Specific steps include cloning the repo and using catkin build.

What's the difference between M-LOAM and LOAM?

M-LOAM extends LOAM by adding online extrinsic calibration for multiple LiDARs, robust fusion, and handling perturbations, while LOAM is designed for single LiDAR odometry and mapping. M-LOAM is more suited for multi-sensor setups.

Does M-LOAM support loop closure?

No, the loop closure functionality is not finished, as mentioned in the system pipeline section. This means it may not handle long-term drift as effectively as other SLAM systems.

What LiDAR sensors are compatible with M-LOAM?

The README doesn't specify exact sensors, but it uses ROS bag datasets, implying compatibility with common LiDARs that output point clouds via ROS topics. Custom setups may require adaptation.

How accurate is the online extrinsic calibration?

Based on the papers cited, it provides accurate calibration validated on real-world datasets like Oxford RoboCar, but accuracy can depend on sensor configurations and environmental conditions.

M-LOAM or Lego-LOAM for multi-sensor robotics?

M-LOAM is better for automated multi-LiDAR calibration and fusion, while Lego-LOAM is optimized for lightweight, ground-vehicle SLAM with single LiDAR. Choose M-LOAM if you need hands-off calibration for multiple sensors.

Open-Awesome

M-LOAM

C++

A robust system for multi-LiDAR extrinsic calibration, real-time odometry, and mapping without manual intervention.

Visit Website GitHub

518 stars95 forks0 contributors

What is M-LOAM?

M-LOAM is a robust system for multi-LiDAR extrinsic calibration, real-time odometry, and mapping. It allows multiple LiDARs with unknown relative poses to automatically calibrate themselves online while providing accurate localization and a globally consistent map. This solves the problem of deploying multi-LiDAR systems without tedious manual calibration.

Target Audience

Robotics researchers and engineers working on autonomous vehicles, drones, or mobile robots that use multiple LiDAR sensors for perception and navigation.

Value Proposition

Developers choose M-LOAM for its fully automated online extrinsic calibration, robust multi-sensor fusion, and real-time performance, eliminating the need for manual calibration and handling hardware perturbations gracefully.

Overview

Robust Odometry and Mapping for Multi-LiDAR Systems with Online Extrinsic Calibration

Use Cases

Best For

Automatically calibrating extrinsic parameters between multiple LiDARs on a vehicle
Real-time odometry and mapping for autonomous driving platforms with multi-LiDAR setups
Robust SLAM in environments where single LiDAR coverage is insufficient
Research on multi-sensor fusion and calibration in robotics
Handling LiDAR sensor failures or extrinsic perturbations in autonomous systems
Building globally consistent maps from multiple uncalibrated LiDAR streams

Not Ideal For

Projects using only a single LiDAR sensor, as it's over-engineered for mono-sensor setups
Teams lacking ROS expertise or unwilling to manage dependencies like Ceres Solver on Ubuntu
Applications requiring complete SLAM with loop closure, since this feature is not finished
Real-time systems on resource-constrained edge devices due to computational demands

Pros & Cons

Pros

Automated Extrinsic Calibration

Enables online calibration of multiple LiDARs without manual intervention, directly addressing the core challenge of multi-sensor setups as stated in the key features.

Real-Time Odometry and Mapping

Provides accurate poses and globally consistent maps in real-time, crucial for autonomous systems, with performance validated on datasets like Simulation Robot and Real Vehicle.

Robust to Perturbations

Maintains functionality against hardware failures or large extrinsic perturbations, as evidenced by the MLOD paper referenced in the README.

Efficient Feature Selection

Uses greedy-based feature selection to reduce algorithm latency, improving efficiency for LiDAR SLAM, as detailed in the related ICRA paper.

Cons

Incomplete Loop Closure

The loop closure part is not finished, limiting its use for applications requiring full SLAM capabilities, as admitted in the system pipeline section.

Complex ROS Dependency

Requires specific Ubuntu versions, ROS installations, and libraries like Ceres Solver, making setup non-trivial and time-consuming for newcomers.

Academic-Focused Implementation

Primarily research-oriented with academic papers, which may lead to less polished documentation and support for production deployment compared to commercial tools.

Frequently Asked Questions

Related Projects

GitHub repository

LIO-SAM: Tightly-coupled Lidar Inertial Odometry via Smoothing and Mapping

Stars4,860

Forks1,521

Last commit1 year ago

pyslam

pySLAM is a hybrid Python/C++ Visual SLAM pipeline supporting monocular, stereo, and RGB-D cameras. It provides a broad set of modern local and global feature extractors, multiple loop-closure strategies, a volumetric reconstruction module, integrated depth-prediction models, and semantic segmentation capabilities for enhanced scene understanding.

Stars3,376

Forks533

Last commit5 days ago

maplab

A Modular and Multi-Modal Mapping Framework

Stars2,857

Forks749

Last commit2 years ago

LeGO-LOAM

LeGO-LOAM: Lightweight and Ground-Optimized Lidar Odometry and Mapping on Variable Terrain

Stars2,745

Forks1,169