xiaojake / ga_slam Goto Github PK

This is an open-source and research-oriented C++ library for solving the SLAM problem on planetary rovers that
aim to navigate and explore extreme terrains. It adopts a globally adaptive technique which uses an a priori global
low-resolution map provided by orbital imagery to enhance the robot's perception and correct its global position.

Created by: Dimitris Geromichalos ([email protected])
Supervised by: Martin Azkarate ([email protected])
Developed at: Planetary Robotics Lab (PRL), European Space Agency (ESA)

• 🌄 Suitable for rough terrain navigation by utilizing 2.5D (elevation) maps
• 🌐 Global pose correction using map matching of local (robot-centric) and global (orbiter) elevation maps
• 📷 Sensor-agnostic data registration using point clouds (support for lidar, stereo camera, ToF camera etc.)
• 💥 Automatic sensor fusion when multiple inputs are provided (without prior configuration)
• 📃 Easily tunable parameter set to fit the needs of different robots and applications

The high-level design diagram shows the main modules that comprise the system.

Although the library can be used on other systems, it was validated using:

• Ubuntu 16.04 (Xenial)
• CMake 3.5.1
• GCC 5.4.0

Required packages:

• Eigen 3 (linear algebra calculations)
• PCL 1.7 (point cloud processing)
• OpenCV 2.4 (image processing)
• Grid Map Core (map structure and calculations)

Optional packages:

• Google Mock (testing)

The dependecies listed above can be installed using this script as well.

To build the library, run:

git clone https://github.com/geromidg/GA_SLAM.git
cd ga_slam
mkdir build && cd build
cmake ..
make

To build and execute the tests, run:

cmake -DENABLE_TESTS=ON ..
make
make test

Make sure to set CMAKE_PREFIX_PATH if you installed the grid_map_core library from the debian package.
For example if you installed it using sudo apt-get install ros-kinetic-grid-map-core, run:

export CMAKE_PREFIX_PATH=$CMAKE_PREFIX_PATH:/opt/ros/kinetic