7. RTAB-Map 3D mapping navigation

wiki: http://wiki.ros.org/rtabmap_ros

7.1. Introduction

This package is a ROS function package for RTAB Map, an RGB-D SLAM method based on a global loop closure detector with real-time constraints. This package can be used to generate 3D point clouds of environments and create 2D occupancy raster maps for navigation.

As can be seen from the above figure, Monte Carlo positioning amcl is not required. RTAB Map has its own positioning function; if used, it will cause repeated positioning and positioning failure. When using RTAB Map to navigate the core framework, the initialized map is provided by RTAB Map, not map_server.

7.2. Map construction and use

Note: When building a map, the slower the speed, the better the effect (note that the rotation speed should be slower). If the speed is too fast, the effect will be poor.

According to different models, you only need to set the purchased model in [.bashrc], X1 (normal four-wheel drive) X3 (Mailun) Take X3 as an example

xxxxxxxxxx
#Raspberry Pi 5 master needs to enter docker first, please perform this step
#If running the script into docker fails, please refer to ROS/07, Docker tutorial
~/run_docker.sh

Open the [.bashrc] file

xxxxxxxxxx
sudo vim .bashrc

Find the [ROBOT_TYPE] parameters and modify the corresponding car model

xxxxxxxxxx
export ROBOT_TYPE=X3 # ROBOT_TYPE: X1 X3 X3plus R2 X7

7.2.1. Start

Start the underlying driver command (robot side)

xxxxxxxxxx
roslaunch yahboomcar_nav laser_astrapro_bringup.launch

<PI5 needs to open another terminal to enter the same docker container

Command to start mapping or navigation (robot side)

xxxxxxxxxx
roslaunch yahboomcar_nav yahboomcar_rtabmap.launch use_rviz:=False

• use_rviz parameter: whether to open rviz.

Start visualization (virtual machine)

xxxxxxxxxx
roslaunch yahboomcar_nav view_rtabmap.launch

Keyboard control node (virtual machine)

xxxxxxxxxx
rosrun teleop_twist_keyboard teleop_twist_keyboard.py # System integration
roslaunch yahboomcar_ctrl yahboom_keyboard.launch # yahboomcar customization

7.2.2. Map construction

After starting according to the above method, choose any method to control the mapping (handle control is recommended); the slower the speed when constructing the map, the better the effect (especially the angular speed); the robot will cover the area to be mapped and the map will be as closed as possible.

• Joystick control

• Keyboard control

When the map construction is completed, directly [ctrl+c] exit the map construction node, and the system will automatically save the map. The default saving path of the map is [~/.ros/rtabmap.db].

View tf tree

xxxxxxxxxx
rosrun rqt_tf_tree rqt_tf_tree

Node view

xxxxxxxxxx
rqt_graph

As can be seen from the above figure, the information that the [rtabmap] node needs to subscribe to: radar data, camera data, and tf data.

7.3. Navigation and obstacle avoidance

Note: [R2] on the remote control handle has the function of canceling the target point.

Start the underlying driver command (robot side)

xxxxxxxxxx
roslaunch yahboomcar_nav laser_astrapro_bringup.launch

Command to start mapping or navigation (robot side)

xxxxxxxxxx
roslaunch yahboomcar_nav yahboomcar_rtabmap_nav.launch use_rviz:=False

• use_rviz parameter: whether to open rviz.

Start visualization (virtual machine)

xxxxxxxxxx
roslaunch yahboomcar_nav view_rtabmap_nav.launch

Keyboard control node (virtual machine)

xxxxxxxxxx
rosrun teleop_twist_keyboard teleop_twist_keyboard.py # System integration
roslaunch yahboomcar_ctrl yahboom_keyboard.launch # yahboomcar customization

When the navigation mode is turned on, the system automatically loads the 2D raster map. It cannot directly load the 3D map and needs to be loaded manually.

Load the three-dimensional map (1, 2, 3), 4 is to add the rviz debugging tool.

At this time, you can manually add [MarkerArray] to facilitate multi-point navigation and observation, and adjust [rviz] display parameters according to needs, such as the size of lidar points.

7.3.1. Single-point navigation

• Use the [2D Pose Estimate] of the [rviz] tool to set the initial pose until the position of the car in the simulation is consistent with the position of the actual car.

• Click [2D Nav Goal] of the [rviz] tool, and then select a target point on the map where there are no obstacles. Release the mouse to start navigation. Only one target point can be selected, and it will stop when it is reached.

7.3.2. Multi-point navigation

• Same as the first step of single-point navigation, first set the initial pose of the car.

• Click [Publish Point] of the [rviz] tool, and then select the target point on the map where there are no obstacles. Release the mouse to start navigation. You can click [Publish Point] again, and then select the point, and the robot will click on it. Cruising between points.

• When using the [2D Pose Estimate] tool of the [rviz] tool to set the initial pose of the car, the multi-point navigation function is automatically canceled.

7.3.3. Parameter configuration

After starting the navigation function, open the dynamic parameter adjustment tool, adjust according to your own needs, and observe the robot's motion status until the effect is optimal. Record the current parameters and modify them to the corresponding dwa_local_planner_params.yaml file under the yahboomcar_nav function package.

xxxxxxxxxx
rosrun rqt_reconfigure rqt_reconfigure

Looking at the yahboomcar_navigation.launch file, we can see that the navigation parameters are modified in the move_base.launch file under the yahboomcar_rtabmap_nav function package.

xxxxxxxxxx
<launch>
     <!-- Whether to open rviz || Whether to open rviz -->
     <arg name="use_rviz" default="false"/>
     <!-- MarkerArray node-->
     <node name='send_mark' pkg="yahboomcar_nav" type="send_mark.py"/>
     <!-- Mobile APP node -->
     <include file="$(find yahboomcar_nav)/launch/library/app.launch"/>
     <!-- Navigation core component move_base -->
     <include file="$(find yahboomcar_nav)/launch/library/move_base.launch"/>
     <!-- rtabmap navigation -->
     <include file="$(find yahboomcar_nav)/launch/library/rtabmap_nav.launch"/>
     <!-- RVIZ -->
     <include file="$(find yahboomcar_nav)/launch/view/view_rtabmap_nav.launch" if="$(arg use_rviz)"/>
</launch>

Find the move_base.launch file and open the example file as follows. It can be modified and replaced according to your needs. At this time, [DWA Planner] is selected and the [DWA] file is loaded.

xxxxxxxxxx
<launch>
     <arg name="robot_type" value="$(env ROBOT_TYPE)" doc="robot_type [X1,X3,X3plus,R2,X7]"/>
     <!-- Arguments -->
     <arg name="move_forward_only" default="false"/>
     <!-- move_base -->
     <node pkg="move_base" type="move_base" respawn="false" name="move_base" output="screen">
         <rosparam file="$(find yahboomcar_nav)/param/common/global_costmap_params.yaml" command="load"/>
         <rosparam file="$(find yahboomcar_nav)/param/common/local_costmap_params.yaml" command="load"/>
         <rosparam file="$(find yahboomcar_nav)/param/common/move_base_params.yaml" command="load"/>
         <rosparam file="$(find yahboomcar_nav)/param/common/costmap_common_params_$(arg robot_type).yaml" command="load"
             ns="global_costmap"/>
         <rosparam file="$(find yahboomcar_nav)/param/common/costmap_common_params_$(arg robot_type).yaml" command="load"
             ns="local_costmap"/>
         <rosparam file="$(find yahboomcar_nav)/param/common/dwa_local_planner_params_$(arg robot_type).yaml" command="load"/>
         <param name="base_local_planner" type="string" value="dwa_local_planner/DWAPlannerROS" if="$(eval arg('robot_type') == 'X3')"/>
         <!-- <param name="base_local_planner" type="string" value="teb_local_planner/TebLocalPlannerROS"/>--><param name="DWAPlannerROS/min_vel_x" value="0.0" if="$(arg move_forward_only)"/>
         <remap from="cmd_vel" to="cmd_vel"/>
         <remap from="odom" to="odom"/>
     </node>
</launch>

Note: When using the DWA planner, the difference between an omnidirectional car and a differential car lies in whether the speed in the Y direction is 0. There are clear comments in it, which can be modified according to the actual situation.

Enter the dwa_local_planner_params.yaml file under the yahboomcar_nav function package. Some parameters are as follows:

xxxxxxxxxx
DWAPlannerROS:
   # Robot Configuration Parameters
   # Absolute value of maximum linear velocity in x direction, unit: meters/second
   # The maximum y velocity for the robot in m/s
   max_vel_x: 0.6
   # The absolute value of the minimum linear velocity in the x direction, a negative number means it can be retreated, unit: meters/second
   # The minimum x velocity for the robot in m/s, negative for backwards motion.
   min_vel_x: -0.6
   # Absolute value of the maximum linear velocity in the y direction, unit: meters/second. The differential robot is 0
   # The maximum y velocity for the robot in m/s
   max_vel_y: 0.3
   # The absolute value of the minimum linear velocity in the y direction, unit: meters/second. The differential robot is 0
   # The minimum y velocity for the robot in m/s
   min_vel_y: -0.3
... ...
   #The ultimate acceleration of the robot in the x direction, the unit is meters/sec^2
   # The x acceleration limit of the robot in meters/sec^2
   acc_lim_x: 10.0
   # The ultimate acceleration of the robot in the y direction, which is 0 for differential robots
   # The y acceleration limit of the robot in meters/sec^2
   acc_lim_y: 10.0
... ..

Other parameter files can be opened, combined with comments and courseware, and modified according to your own needs.

7.4, node rtabmap

This is the master node for this package. It is a wrapper around the RTAB mapping core library. Here, when loop closure is detected, the map is incrementally built and optimized. The node's online output is this map, which contains the latest data added to the map. The default location of the RTAB map database is [.ros/rtabmap.db], and the workspace is also set to [.ros].

7.4.1. Subscribe to topics

7.4.2. Publishing topics

7.4.3, Service

7.4.4. Parameters

7.4.5, tf conversion

what is needed:

• base_link → sensor

• odom → base_link

which provided:

map → odom

7.5, node rtabmapviz

This node starts the visual interface of RTAB-Map. It is a wrapper for the RTAB-MapGUI library. Its purpose is the same as rviz, but with specific options for RTAB-Map.

7.5.1. Subscribe to topics

7.5.2. Parameter configuration

7.5.3. Required tf conversion

• base_link → sensor coordinate system

• odom → base_link

• map → odom