GMapping-SLAM Mapping

1. Course Content

Basic: Run the sample program to perform SLAM mapping using the GMapping algorithm and save the grid map.

Advanced: Understand the configuration parameters of the GMapping algorithm.

2. GMapping Algorithm Introduction

• GMapping is only suitable for 2D laser scans with fewer than 1440 points per frame. If the number of points exceeds 1440, an issue like [mapping-4] process has died will occur.
• GMapping is a popular open-source SLAM algorithm based on the filter SLAM framework.
• GMapping is based on the RBPF (Rao-Blackwellized Particle Filter) algorithm, where the localization and mapping processes are separated. Localization is performed first, followed by mapping.
• GMapping makes two main improvements to the RBPF algorithm: an improved proposal distribution and selective resampling.

Advantages: GMapping can build indoor maps in real-time. It requires less computational power and offers higher accuracy for small-scale mapping.

Disadvantages: As the scene size increases, more particles are needed. Since each particle carries a map, both memory and computational requirements increase when building large maps. Therefore, it is not suitable for large-scale mapping. It also lacks loop closure detection, which can cause map misalignment when a loop is closed. Although increasing the number of particles can help close the map, it comes at the cost of increased computation and memory.

3. Start the Agent

Note: If it is already running, you do not need to start it again.

Enter the command in the vehicle's terminal:

start_agent

The terminal will print the following information, indicating a successful connection:

4. Run the Example

4.1 Mapping Process

• Note: When mapping, a slower speed (especially rotational speed) yields better results. If the speed is too high, the mapping quality will be poor.
• Launch the mapping command in the terminal:

For Raspberry Pi 5 users, if the ROS container is not started, you need to start it first,

xxxxxxxxxx
bringup_m3

Open a terminal inside the container,

xxxxxxxxxx
exec_m3

Enter the command,

(ORIN & PI5 boards)

xxxxxxxxxx
ros2 launch m3_bringup gmapping.launch.py

(RDK board) Run in separate steps, with visualization launched on the accompanying virtual machine Rosmaster-M3-Ubuntu22.04,

xxxxxxxxxx
# On the virtual machine
rviz2 -d /home/yahboom/yahboomM3_ws/user_ws/src/m3_bringup/rviz/gmapping_rviz.rviz
# On the host machine
ros2 launch m3_bringup gmapping.launch.py gui:=False

• To control the vehicle's movement, you can use either keyboard or joystick control. Here, we use keyboard control as an example:

xxxxxxxxxx
keyboard_control

• Click on the terminal window with the mouse and press z to reduce the speed appropriately.

Press I, <, J, L to control the car to move forward, backward, turn left, and turn right, respectively, to complete the mapping.

4.2 Save the Map

• Run the following command in the terminal to save the grid map:

xxxxxxxxxx
ros2 launch m3_bringup save_map.launch.py

[!TIP]

Optional Parameters:

map_name: Specifies the name of the saved map, with the default value being map.

For example, to save a map named test.yaml, add the parameter: ros2 launch m3_bringup save_map.launch.py map_name:=test

The terminal prompt Map saved successfully indicates that the map was saved successfully. If it fails, try saving again.

The default map save path is as follows:

xxxxxxxxxx
~/map/map.yaml

This includes a PGM image and a YAML file, yahboom_map.yaml.

xxxxxxxxxx
image: yahboom_map.pgm
mode: trinary
resolution: 0.05
origin: [-10, -10, 0]
negate: 0
occupied_thresh: 0.65
free_thresh: 0.25

Parameter analysis:

• image: The path to the map file, which can be an absolute or relative path.
• mode: This attribute can be trinary, scale, or raw, depending on the chosen mode. trinary is the default.
• resolution: The resolution of the map, in meters/pixel.
• origin: The 2D pose (x, y, yaw) of the bottom-left corner of the map. Here, yaw is a counter-clockwise rotation (yaw=0 means no rotation). Many parts of the system currently ignore the yaw value.
• negate: Whether to invert the meaning of white/black and free/occupied (the interpretation of thresholds is not affected).
• occupied_thresh: Pixels with an occupancy probability greater than this threshold are considered fully occupied.
• free_thresh: Pixels with an occupancy probability less than this threshold are considered fully free.

5. Node Analysis

5.1 Display Node Computation Graph

Run in the virtual machine terminal:

xxxxxxxxxx
ros2 run rqt_graph rqt_graph

5.2 TF Transformation

Run in the virtual machine terminal:

xxxxxxxxxx
ros2 run rqt_tf_tree rqt_tf_tree 

The image size is too large. The original image can be viewed in this lesson's course folder.

5.3 GMapping Node Details

xxxxxxxxxx
ros2 node info /slam_gmapping

Entering the above command in the terminal allows you to view the topics subscribed to and published by the GMapping node.

​x
/slam_gmapping
  Subscribers:
    /parameter_events: rcl_interfaces/msg/ParameterEvent
    /scan: sensor_msgs/msg/LaserScan
  Publishers:
    /entropy: std_msgs/msg/Float64
    /map: nav_msgs/msg/OccupancyGrid
    /map_metadata: nav_msgs/msg/MapMetaData
    /parameter_events: rcl_interfaces/msg/ParameterEvent
    /rosout: rcl_interfaces/msg/Log
    /tf: tf2_msgs/msg/TFMessage
  Service Servers:
    /slam_gmapping/describe_parameters: rcl_interfaces/srv/DescribeParameters
    /slam_gmapping/get_parameter_types: rcl_interfaces/srv/GetParameterTypes
    /slam_gmapping/get_parameters: rcl_interfaces/srv/GetParameters
    /slam_gmapping/list_parameters: rcl_interfaces/srv/ListParameters
    /slam_gmapping/set_parameters: rcl_interfaces/srv/SetParameters
    /slam_gmapping/set_parameters_atomically: rcl_interfaces/srv/SetParametersAtomically
  Service Clients:
​
  Action Servers:
​
  Action Clients:

5.4 Configuration File

• Configuration file path:

xxxxxxxxxx
~/yahboomM3_ws/user_ws/src/m3_bringup/config/slam_gmapping.yaml

• Default configuration parameters:

xxxxxxxxxx
slam_gmapping:
  ros__parameters:
    # Angular update threshold (radians): Update the map when the robot rotates more than this angle
    angularUpdate: 0.25
    # Angular sampling step (radians): Angular search step for scan matching
    astep: 0.05
    # Robot base coordinate frame name
    base_frame: base_footprint
    # Map coordinate frame name
    map_frame: map
    # Odometry coordinate frame name
    odom_frame: odom
    # Iteration termination condition for scan matching: Stop iterating when parameter change is less than this value
    delta: 0.05
    # Maximum number of iterations for scan matching
    iterations: 5
    # Kernel size for scan matching, affects matching smoothness
    kernelSize: 1
    
    # Distance sampling range for large-scale scan matching (meters)
    lasamplerange: 0.005
    # Distance sampling step for large-scale scan matching (meters)
    lasamplestep: 0.005
    # Linear update threshold (meters): Update the map when the robot moves more than this distance
    linearUpdate: 0.5
    # Distance sampling range for small-scale scan matching (meters)
    llsamplerange: 0.01
    # Distance sampling step for small-scale scan matching (meters)
    llsamplestep: 0.01
    
    # Measurement noise covariance for laser scan
    lsigma: 0.075
    # Number of laser beams to skip: 0 means use all laser points
    lskip: 0
    # Linear sampling step (meters): Linear search step for scan matching
    lstep: 0.05
    # Map update time interval (seconds)
    map_update_interval: 3.0
    # Maximum laser range (meters): Points beyond this distance are ignored
    maxRange: 6.0
    # Maximum usable laser range (meters): Maximum distance used for mapping
    maxUrange: 4.0
    # Minimum score threshold for scan matching: Matches below this score are discarded
    minimum_score: 0.0
    # Occupancy probability threshold: Cells with a probability greater than this are considered occupied
    occ_thresh: 0.25
    # Gain factor: Affects the speed of map updates
    ogain: 3.0
    # Number of particles for the particle filter
    particles: 30
    # QoS (Quality of Service) parameter settings
    qos_overrides:
      /parameter_events:
        publisher:
          depth: 1000          # Message queue depth
          durability: volatile # Durability policy
          history: keep_all    # History policy
          reliability: reliable # Reliability policy
      /tf:
        publisher:
          depth: 1000
          durability: volatile
          history: keep_last   # Keep only the latest message
          reliability: reliable
    # Resampling threshold: Trigger resampling when the number of effective particles is below this ratio
    resampleThreshold: 0.5
    # Angular noise covariance for scan matching
    sigma: 0.05
    # Odometry model parameter: Proportional linear error from linear motion
    srr: 0.1
    # Odometry model parameter: Proportional angular error from linear motion
    srt: 0.2
    # Odometry model parameter: Proportional linear error from angular motion
    str: 0.1
    # Odometry model parameter: Proportional angular error from angular motion
    stt: 0.2
    # Time update interval (seconds): Update the map even if the robot has not moved
    temporalUpdate: 1.0
    # Coordinate transform publication period (seconds)
    transform_publish_period: 0.05
    # Whether to use simulation time
    use_sim_time: false
    # Maximum x-range of the map (meters)
    xmax: 100.0
    # Minimum x-range of the map (meters)
    xmin: -100.0
    # Maximum y-range of the map (meters)
    ymax: 100.0
    # Minimum y-range of the map (meters)
    ymin: -100.0

These are the configurable parameters for GMapping. If you need to modify the configuration parameters, you must recompile the slam_gmapping package in the M3Pro_ws workspace for the changes to take effect:

xxxxxxxxxx
colcon build --packages-select slam_gmapping