How to set up Elevation Mapping with a new robot?

Start by duplicating and adapting parameter files from the demos package, like those in the 'config' folder, and modify the launch file to match your robot's sensors and TF frames. The README recommends beginning with the simple_demo example for realistic deployment.

Elevation Mapping vs gmapping: which is better for terrain?

Elevation Mapping focuses on local elevation maps with uncertainty handling for rough terrain navigation, while gmapping is for 2D SLAM with laser scanners. Choose Elevation Mapping if you need probabilistic height data and pose drift robustness.

Can Elevation Mapping work with stereo cameras?

Yes, it supports stereo cameras as distance sensors, listed under multi-sensor support. Configure the input source as a pointcloud topic from the stereo output in the parameters.

How does Elevation Mapping handle pose uncertainty?

It aggregates uncertainty through robot motion using a probabilistic model, fusing sensor data based on Mahalanobis distance thresholds to account for localization drift, as explained in the methodology section.

Is Elevation Mapping compatible with ROS 2?

No, it's built for ROS and is unmaintained, so it likely doesn't support ROS 2 natively. Porting would require significant effort due to ROS 1 dependencies and lack of updates.

How to tune the Mahalanobis distance threshold?

Adjust the threshold in parameters to balance data fusion: lower values make the map conservative, while higher values adapt faster to new measurements. The default is 2.5, but tune based on sensor noise and environment dynamics.

Open-Awesome

elevation_mapping

BSD-3-ClauseC++

A ROS package for robot-centric elevation mapping that handles pose uncertainty for navigation on rough terrain.

GitHub

1.8k stars508 forks0 contributors

What is elevation_mapping?

Elevation Mapping is a ROS package that creates local elevation maps for mobile robots navigating rough terrain. It uses robot-centric mapping to handle pose estimation drift by aggregating uncertainty through the robot's motion, providing a probabilistic terrain representation. The package processes distance sensor data (e.g., from LiDAR or depth cameras) to generate elevation maps that are essential for locomotion planning and local navigation.

Target Audience

Robotics researchers and engineers working on mobile robot navigation, particularly those using legged robots or ground vehicles in unstructured outdoor environments. It's designed for teams with ROS expertise who need reliable terrain perception for autonomous locomotion.

Value Proposition

Developers choose Elevation Mapping for its explicit handling of pose uncertainty, which is critical for robots operating in GPS-denied or slippery terrain. Its robot-centric approach and probabilistic fusion provide more reliable maps than traditional methods when localization drifts, and its tight ROS integration makes it easy to incorporate into existing autonomy stacks.

Overview

Robot-centric elevation mapping for rough terrain navigation

Use Cases

Best For

Legged robot locomotion planning over rough terrain
Mobile robot navigation in unstructured outdoor environments
Terrain mapping with uncertain robot localization
Real-time elevation mapping for autonomous vehicles
Research on probabilistic terrain perception
Integrating multiple depth sensors for robust terrain estimation

Not Ideal For

Projects requiring long-term maintenance and active support, as the package is no longer maintained
Applications needing global, consistent maps over large static areas, since it focuses on local robot-centric mapping
Teams not using ROS or with limited ROS expertise, due to tight integration and complex dependencies
Simple indoor navigation with precise localization, where the probabilistic overhead is unnecessary

Pros & Cons

Pros

Pose Drift Robustness

Explicitly handles localization uncertainty by maintaining a robot-centric map that aggregates drift, crucial for navigation in rough terrain as described in the cited papers.

Probabilistic Data Fusion

Uses Mahalanobis distance thresholds to fuse sensor data probabilistically, providing reliable height estimates even with noisy inputs, as detailed in the parameters section.

Multi-Sensor Flexibility

Supports various distance sensors like Kinect and LiDAR through configurable input sources, demonstrated in the parameter examples for pointcloud topics.

Dynamic Environment Adaptation

Includes visibility cleanup and continuous cleanup options to manage moving objects and outdated measurements, enhancing map accuracy in changing terrains.

Cons

Unmaintained Codebase

The README notes it is no longer actively maintained, risking compatibility issues with newer ROS versions and lack of bug fixes or support.

Complex Setup Requirements

Depends on multiple ROS packages and libraries like Grid Map, kindr, and PCL, making initial installation and integration challenging for new users.

Research Code Instability

Advertised as research code with frequent changes and no fitness guarantee, making it unreliable for production deployments without significant customization.

Frequently Asked Questions

Related Projects

GitHub repository

Open3D: A Modern Library for 3D Data Processing

Draco is a library for compressing and decompressing 3D geometric meshes and point clouds. It is intended to improve the storage and transmission of 3D graphics.

Stars7,415

Forks1,063

Last commit22 days ago

mmdetection3d

OpenMMLab's next-generation platform for general 3D object detection.

Stars6,486

Forks1,784

Last commit2 years ago

OpenPCDet

OpenPCDet Toolbox for LiDAR-based 3D Object Detection.

Stars5,660

Forks1,461

Last commit9 months ago

Community-curated · Updated weekly · 100% open source

Found a gem we're missing?

Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.

Submit a project Star on GitHub