Seva's (on the left) and Enrique's (on the right) Tree Inspection Robot.
The project is a final project for the IFROS third semester's subject "IFRoS Lab" in ELTE University, Budapest.
- Follow README from artwifr_gazebo to use robot in simulation;
- Install kiss_icp_ros package for exploration and mapping:
cd ~/catkin_ws/src
git clone https://github.com/sevagul/kiss-icp-ros.git
Follow it's README to install it's dependencies, afterwards build the workspace.
The kiss-icp may be configured in the param/icp_params.yaml file.
- Install velodyne_height_map package for mapping;
- Install frontier_exploration package for exploration;
To install all the dependencies, use
cd <your/workspace/path>
rosdep install --from-paths src --ignore-src -r -y
And finally build everything using catkin build or catkin_make
To run the simulation with the groundtruth odometry:
roslaunch atwifr_bringup simulation.launch
To run the simulation with the kiss-icp odometry:
roslaunch atwifr_bringup sim_with_icp.launch
Before launching mapping, be sure that all the odometry transformations are being broadcasted.
After simulation/robot is running, launch mapping with
roslaunch atwifr_bringup octomap.launch
To manually navigate the robot, run
rosrun teleop_twist_keyboard teleop_twist_keyboard.py
To setup the navigation, run
roslaunch atwifr_bringup navigation.launch
You can specify a waypoint from rviz to test how it works
If you want to run it with kiss-icp, you may need to run the following before:
rosrun atwifr_bringup odom_from_tf
This node will generate odometry messages from the /tf provided by the kiss-icp node. It uses some filtering so that the speed provided in the odometry is a bit smoothed.
Other odometry nodes may broadcast odometry messages themselves, so there might be no need to run it.
For exploration, run the following:
roslaunch frontier_exploration explore_costmap.launch costmap:=projected_map
To run the inspection, run
rosrun atwifr_bringup tsp
and call the service when ready to start inspection as the following:
rosservice call /plan_tsp
Note, that for now inspection works best when simulation is launched with groundtruth instead of kiss-icp.
This module is not integrated with the ones above, and can be read, installed and used from the following repository
In the next figure, you can see the inspection task in progress
The inspection points are marked with the red boxes, the tsp path in drawn with the dark line
You can see a video of the inspection process here
In the following Figure you can see the explored map (on the left) and the exploration in progress (on the right):
The detected frontier is marked with the blue color
And here is a small demo video
KISS-ICP odometry was used to produce these results.
You can also see the example of the pointclouds aligned by the kiss-icp in kiss-icp-ros repository
The examples of TREE SLAM execution can be seen in the TREE SLAM repository