LRAE

LRAE is a Large-Region-Aware Exploration method that enables a ground robot to autonomously explore uneven terrains.

To obtain high exploration efficiency, LRAE adopts a large-region-aware exploration route optimization strategy that prioritizes exploring large regions while also considering exploring nearby small regions. To safely and completely explore uneven terrains, LRAE fully introduces traversability information to extract unknown regions and assess exploration safety levels. The safety levels are then integrated into the design of the exploration strategy to ensure safe robotic exploration. We validate LRAE in various challenging simulation scenes and real-world wild uneven terrains. The results show that our method can safely explore uneven terrains and improve exploration efficiency by up to 45.3% compared with state-of-the-art methods.

Video Link: Video on Youtube; Video on Bilibili

Related Paper: Paper on IEEE

Q. Bi, X. Zhang, S. Zhang, R. Wang, L. Li and J. Yuan, "LRAE: Large-Region-Aware Safe and Fast Autonomous Exploration of Ground Robots for Uneven Terrains," in IEEE Robotics and Automation Letters, vol. 9, no. 12, pp. 11186-11193.

(If it is useful to you, please cite our paper and ⭐️ our code.)

Other Links: multi-robot exploration code CURE1

Prerequisites

1. LRAE has been tested on Ubuntu 20.04 with ROS Noetic, please run the following commands to install required dependencies or tools:

sudo apt-get install ros-noetic-gazebo-* \
ros-noetic-gazebo-ros-control* \
ros-noetic-controller-* \
ros-noetic-ros-controllers \
ros-noetic-ros-control \
ros-noetic-tf2-* \
ros-noetic-velodyne-* \
ros-noetic-robot-state-publisher* \
ros-noetic-joint-state-controller* \
ros-noetic-velocity-controllers* 

2. In addition, we recommend that you download gazebo_models and Supplementary Gazebo Models for LRAE to the directory ~/.gazebo/models.

Build LRAE

Then simply clone and compile our package:

cd ${YOUR_WORKSPACE_PATH}/src
git clone [email protected]:NKU-MobFly-Robotics/LRAE.git
cd ../ 
catkin_make

Known issues

1. resource not found: velodyne_description
   ROS path [0]=/opt/ros/noetic/share/ros
   ROS path [1]=/${YOUR_WORKSPACE_PATH}/src
   ROS path [2]=/opt/ros/noetic/share
   

   sudo apt-get install ros-noetic-velodyne-*

2. E: Unable to locate package ros-noetic-velodyne-*
   E: Couldn't find any package by glob 'ros-noetic-velodyne-*'
   E: Couldn't find any package by regex 'ros-noetic-velodyne-*'
   

   sudo apt-get update

3. [INFO] [1724678398.723761, 1733.140000]: Loading controller: scout_motor_rr_controller
   [ERROR] [1724678398.729870298, 1733.145000000]: Could not load controller 'scout_motor_rr_controller' because controller type 'velocity_controllers/JointVelocityController' does not exist.
   [ERROR] [1724678398.729950407, 1733.146000000]: Use 'rosservice call controller_manager/list_controller_types' to get the available types
   [ERROR] [1724678399.731316, 1733.943000]: Failed to load scout_motor_rr_controller
   

   sudo apt-get install ros-noetic-ros-controllers ros-noetic-ros-control

4. Some common issues can also be referred to on the "Issues" page.

Run LRAE in simulation

After the compilation is successful, you need to open two terminals and run the following two commands.

source devel/setup.bash && roslaunch fitplane simulation_scene1.launch

source devel/setup.bash && roslaunch lrae_planner exploration_scene1.launch

After running the above two commands, if you can see the red route path, the purple exploration path, the green local path, and the robot has begun moving to explore in the RVIZ interface, it means that LRAE has started successfully.

Run LRAE in different scenes

We have designed four testing scenes with different characteristics and sizes：

Scene 1 Scene 2

If you want to run LRAE in Scene 2, you need to modify the above two commands to：

source devel/setup.bash && roslaunch fitplane simulation_scene2.launch

source devel/setup.bash && roslaunch lrae_planner exploration_scene2.launch

For Scene 3 and Scene 4, the method is the same as above.

Run LRAE in other scenes

Schematic diagram of some parameters:

We assume that exploration problems have boundaries, otherwise, exploration will continue indefinitely. Therefore, you need first to define the exploration boundary for the robot according to the scene, then set the parameters of the globalMapData according to the exploration boundary to ensure that the range of the exploration boundary is within the range of the globalMapData.

Please follow the following steps:

Let the robot's initial position be the coordinate origin, the robot's orientation be the positive direction of the x-axis, and the y-axis follows the right-hand coordinate system, the range of the exploration boundary in this coordinate system should be within the range of globalMapData determined by the four parameters map_w, map_h, mapinitox, and mapinitoy in node “exploration_map_merge”:

1. Add the following parameters to node “Traversibility_mapping”：

​<param name="use_ex_range" value="true"/>
<param name="ex_robot_back" value="-10.0"/>
​<param name="ex_robot_right" value="-10.0"/>
​<param name="ex_robot_front" value="50.0"/>
​<param name="ex_robot_left" value="50.0"/>

2. Modify the following parameters of the globalMapData in node “exploration_map_merge”：

​<param name="map_w" type="int" value="200" />
​<param name="map_h" type="int" value="200" />
​<param name="mapinitox" type="double" value="-10.0" />
​<param name="mapinitoy" type="double" value="-10.0" />

3. The conditions that need to be met between parameters:
   1. ​If it is necessary to define the exploration boundary, use_ex_range is true; otherwise, it is false;
   2. ​ex_robot_front represents the farthest distance that can be explored along the positive x-axis；
   3. ​ex_robot_back represents the farthest distance that can be explored along the negative x-axis；
   4. ​ex_robot_left represents the farthest distance that can be explored along the positive y-axis;
   5. ​ex_robot_right represents the farthest distance that can be explored along the negative y-axis;
   6. ​map_w is greater than or equal to ((ex_robot_front + abs(ex_robot_back)) / map resolution) then round up
   7. ​map_h is greater than or equal to ((ex_robot_left + abs(ex_robot_right)) / map resolution) then round up;
   8. ​note: map resolution has been set to 0.3 in this code repository；
   9. ​mapinitox is less than or equal to ex_robot_back;
   10. ​mapinitoy is less than or equal to ex_robot_right;
   11. ​In conclusion, the exploration boundary range defined by the “Traversibility_mapping” node must be entirely within the range of the globalMapData determined by the “exploration_map_merge” node.

Main Parameters

Main parameters affecting terrain traversability analysis:

float max_angle_ = 40.0;
float max_flatness_ = 0.01;
float w1_ = 0.8;

Main parameters affecting exploration performance:

<param name="angle_pen" type="double" value ="0.45" />
<param name="update_cen_thre" type="int" value="6" />
<param name="unknown_num_thre" type="int" value ="200" />
<param name="minrange" type="double" value="20.0" />
<param name="limit_max_square" type="bool" value ="true" />
<param name="use_go_end_nearest" type="bool" value="true" />
<param name="end_neacen_disthre" type="double" value ="10.0" />
<param name="end_cur_disrate" type="double" value="2.0" />

If the robot ends up exploring some tiny unknown regions back and forth, please increase the unknown_num_thre parameter appropriately. Remark: Since the Real Time Factor in Gazebo is not always equal to 1, please use the ros::Time class to measure the exploration time instead of using the actual wall-clock time.

Acknowledgements

We sincerely appreciate the following open source projects: FAEL, TARE, PUTN, and Ji Zhang's local_planner.