ROS2 package port for multi-robot autonomous exploration of [m-explore](https://github.com/hrnr/m-explore). Currently tested on Eloquent, Dashing, Foxy, and Galactic distros.
You can open rviz2 and add the exploration frontiers marker (topic is `explore/frontiers`) to see the algorithm working and the frontier chosen to explore.
### Additional features
#### Stop/Resume exploration
By default the exploration node will start right away the frontier-based exploration algorithm. Alternatively, you can stop the exploration by publishing to a `False` to `explore/resume` topic. This will stop the exploration and the robot will stop moving. You can resume the exploration by publishing to `True` to `explore/resume`.
This package works with known and unknown initial poses of the robots. It merges the maps of the robots and publishes the merged map. Some results in simulation are shown below.
This modality gives normally the best results. The original ROS1 code only supports [slam_gmapping](https://github.com/ros-perception/slam_gmapping) type of maps for the merge. The following shows the result with that.
We also support using [slam_toolbox](https://github.com/SteveMacenski/slam_toolbox) in a yet [experimental branch](https://github.com/robo-friends/m-explore-ros2/tree/feature/slam_toolbox_compat). The following demo shows the map merging using the currently supported and most used ROS2-SLAM library.
Because of the logic that merges the maps, currently as a straightforward port to ROS2 from the ROS1 version, the SLAM needs to be done using the ROS1 defacto slam option which is [slam_gmapping](https://github.com/ros-perception/slam_gmapping), which hasn't been ported officially to ROS2 yet. There is an unofficial port but it lacks to pass a namespace to its launch file. For that, this repo was tested with one of the authors of this package's [fork](https://github.com/charlielito/slam_gmapping/tree/feature/namespace_launch). You'll need to git clone to your workspace and build it with colcon.
**Note**: You could use [slam_toolbox](https://github.com/SteveMacenski/slam_toolbox) instead but you need to use this [experimental branch](https://github.com/robo-friends/m-explore-ros2/tree/feature/slam_toolbox_compat) which is still under development.
To spawn multiple robots, you need the `nav2_gazebo_spawner` which does not come up with the `nav2-bringup` installation. For that, install it with `sudo apt install ros-${ROS_DISTRO}-nav2-gazebo-spawner`.
Note that was the case for release previous to `humble` but since `humble` release, this package is deprecated and a gazebo node is used for this. So, if you are using `humble` or newer, you don't need to install it.
This repo has some config examples and launch files for running this package with 2 TB3 robots and a world with nav2. Nonetheless, they are only compatible with the galactic/humble distros and since some breaking changes were introduced in this distro, if you want to try it with another ros2 distro you'll need to tweak those param files for that nav2's distro version (which shouldn't be hard).
By default, the demo runs with known initial poses. You can change that by launching again both launch commands with the flag `known_init_poses:=False`
**Note**: If you want to use slam_toolbox, launch `multirobot_map_merge` with the following flag instead: `slam_toolbox:=True`. Remember to use the experimental branch mentioned above.
