Skip to main content

​
Overview

Google Cartographer is an advanced SLAM system developed by Google. It provides real-time 2D and 3D mapping with sophisticated loop closure and optimization.
Cartographer vs SLAM Toolbox: Cartographer excels in complex, multi-floor environments. SLAM Toolbox is simpler and works well for single-floor indoor spaces.

​
When to Use Cartographer

Use Cartographer

  • Multi-floor buildings
  • Large outdoor areas
  • 3D mapping (with 3D LiDAR)
  • Complex loop closures
  • High-accuracy requirements

Use SLAM Toolbox

  • Single-floor indoor
  • Simple environments
  • Limited compute (Raspberry Pi)
  • Quick setup needed
  • Real-time constraints

​
Installation

# Install Cartographer
sudo apt install ros-jazzy-cartographer ros-jazzy-cartographer-ros
Check installation: 
ros2 pkg list | grep cartographer
# Expected:
# cartographer
# cartographer_ros
# cartographer_ros_msgs

​
Configuration

​
Configuration File

Cartographer uses Lua configuration files. File: config/cartographer.lua 
include "map_builder.lua"
include "trajectory_builder.lua"

options = {
  map_builder = MAP_BUILDER,
  trajectory_builder = TRAJECTORY_BUILDER,

  -- Frame IDs
  map_frame = "map",
  tracking_frame = "base_link",
  published_frame = "base_link",
  odom_frame = "odom",
  provide_odom_frame = true,
  publish_frame_projected_to_2d = true,
  use_pose_extrapolator = true,

  -- Odometry
  use_odometry = true,

  -- Sensors
  use_nav_sat = false,  -- No GPS
  use_landmarks = false,

  num_laser_scans = 1,  -- Single LiDAR
  num_multi_echo_laser_scans = 0,
  num_subdivisions_per_laser_scan = 1,
  num_point_clouds = 0,  -- 2D only

  -- Lookup transforms
  lookup_transform_timeout_sec = 0.2,
  submap_publish_period_sec = 0.3,
  pose_publish_period_sec = 5e-3,  # 200Hz
  trajectory_publish_period_sec = 30e-3,
  rangefinder_sampling_ratio = 1.0,
  odometry_sampling_ratio = 1.0,
  fixed_frame_pose_sampling_ratio = 1.0,
  imu_sampling_ratio = 1.0,
  landmarks_sampling_ratio = 1.0,
}

-- 2D SLAM Configuration
MAP_BUILDER.use_trajectory_builder_2d = true
TRAJECTORY_BUILDER_2D.use_imu_data = true
TRAJECTORY_BUILDER_2D.min_range = 0.15  -- RPLIDAR A1M8 min
TRAJECTORY_BUILDER_2D.max_range = 12.0  -- RPLIDAR A1M8 max
TRAJECTORY_BUILDER_2D.missing_data_ray_length = 5.0
TRAJECTORY_BUILDER_2D.use_online_correlative_scan_matching = true
TRAJECTORY_BUILDER_2D.ceres_scan_matcher.translation_weight = 10
TRAJECTORY_BUILDER_2D.ceres_scan_matcher.rotation_weight = 40

-- Submaps
TRAJECTORY_BUILDER_2D.submaps.num_range_data = 90
TRAJECTORY_BUILDER_2D.submaps.grid_options_2d.resolution = 0.05  -- 5cm

-- Motion filter (when to add scans)
TRAJECTORY_BUILDER_2D.motion_filter.max_time_seconds = 5.0
TRAJECTORY_BUILDER_2D.motion_filter.max_distance_meters = 0.2
TRAJECTORY_BUILDER_2D.motion_filter.max_angle_radians = math.rad(10)

-- Loop closure
POSE_GRAPH.optimize_every_n_nodes = 90
POSE_GRAPH.constraint_builder.min_score = 0.55
POSE_GRAPH.constraint_builder.global_localization_min_score = 0.60

return options

​
Launch File

File: launch/cartographer.launch.py 
from launch import LaunchDescription
from launch.actions import DeclareLaunchArgument, IncludeLaunchDescription
from launch.substitutions import LaunchConfiguration, PathJoinSubstitution
from launch_ros.actions import Node
from launch_ros.substitutions import FindPackageShare

def generate_launch_description():
    # Paths
    pkg_share = FindPackageShare('mecanum_description').find('mecanum_description')
    cartographer_config_dir = PathJoinSubstitution([pkg_share, 'config'])
    configuration_basename = 'cartographer.lua'

    # Arguments
    resolution = LaunchConfiguration('resolution')
    publish_period_sec = LaunchConfiguration('publish_period_sec')

    declare_resolution_argument = DeclareLaunchArgument(
        'resolution',
        default_value='0.05',
        description='Resolution of a grid cell in the published occupancy grid')

    declare_publish_period_sec_argument = DeclareLaunchArgument(
        'publish_period_sec',
        default_value='1.0',
        description='OccupancyGrid publishing period')

    # Cartographer node
    start_cartographer_node = Node(
        package='cartographer_ros',
        executable='cartographer_node',
        name='cartographer_node',
        output='screen',
        parameters=[{'use_sim_time': False}],
        arguments=[
            '-configuration_directory', cartographer_config_dir,
            '-configuration_basename', configuration_basename
        ],
        remappings=[
            ('scan', '/scan'),
            ('odom', '/odometry/filtered'),
            ('imu', '/imu/data')
        ]
    )

    # Occupancy grid node
    start_cartographer_occupancy_grid_node = Node(
        package='cartographer_ros',
        executable='cartographer_occupancy_grid_node',
        name='cartographer_occupancy_grid_node',
        output='screen',
        parameters=[
            {'use_sim_time': False},
            {'resolution': resolution},
            {'publish_period_sec': publish_period_sec}
        ]
    )

    return LaunchDescription([
        declare_resolution_argument,
        declare_publish_period_sec_argument,
        start_cartographer_node,
        start_cartographer_occupancy_grid_node,
    ])

​
Usage

​
Launch Cartographer

# Launch robot control + localization + LiDAR
ros2 launch mecanum_description robot_full.launch.py

# Launch Cartographer
ros2 launch mecanum_description cartographer.launch.py

​
RViz Visualization

rviz2
Configuration:
  1. Fixed Frame: map
  2. Add → Map: Topic /map
  3. Add → LaserScan: Topic /scan
  4. Add → RobotModel
  5. Add → TF
Optionally add submaps:
  • Add → Map: Topic /submap_list (shows Cartographer’s internal submaps)

​
Drive and Map

Use teleop to drive robot: 
ros2 run teleop_twist_keyboard teleop_twist_keyboard
Mapping tips:
  • Moderate speed: 0.3-0.5 m/s
  • Smooth motions (avoid sudden stops)
  • Revisit areas for loop closure
  • Cover entire environment systematically

​
Saving Maps

​
Finish Trajectory

When mapping complete: 
ros2 service call /cartographer/finish_trajectory \
  cartographer_ros_msgs/srv/FinishTrajectory \
  "{trajectory_id: 0}"

​
Save State

Save Cartographer state (for later continuation): 
ros2 service call /cartographer/write_state \
  cartographer_ros_msgs/srv/WriteState \
  "{filename: '/home/user/maps/cartographer_state.pbstream'}"
File: .pbstream (protobuf stream) contains full SLAM state

​
Export to PGM

Convert to occupancy grid map: 
ros2 run cartographer_ros cartographer_pbstream_to_ros_map \
  -pbstream_filename=/home/user/maps/cartographer_state.pbstream \
  -map_filestem=/home/user/maps/my_map
Creates:
  • my_map.pgm
  • my_map.yaml

​
Pure Localization Mode

Localize in existing map:

​
Localization Config

File: config/cartographer_localization.lua 
include "map_builder.lua"
include "trajectory_builder.lua"

options = {
  map_builder = MAP_BUILDER,
  trajectory_builder = TRAJECTORY_BUILDER,

  map_frame = "map",
  tracking_frame = "base_link",
  published_frame = "base_link",
  odom_frame = "odom",

  provide_odom_frame = true,
  publish_frame_projected_to_2d = true,
  use_pose_extrapolator = true,

  use_odometry = true,
  use_nav_sat = false,
  use_landmarks = false,

  num_laser_scans = 1,
  num_multi_echo_laser_scans = 0,
  num_subdivisions_per_laser_scan = 1,
  num_point_clouds = 0,

  lookup_transform_timeout_sec = 0.2,
  submap_publish_period_sec = 0.3,
  pose_publish_period_sec = 5e-3,
  trajectory_publish_period_sec = 30e-3,
  rangefinder_sampling_ratio = 1.0,
  odometry_sampling_ratio = 1.0,
  fixed_frame_pose_sampling_ratio = 1.0,
  imu_sampling_ratio = 1.0,
  landmarks_sampling_ratio = 1.0,
}

-- Pure localization (no mapping)
MAP_BUILDER.use_trajectory_builder_2d = true
TRAJECTORY_BUILDER_2D.use_imu_data = true
TRAJECTORY_BUILDER_2D.min_range = 0.15
TRAJECTORY_BUILDER_2D.max_range = 12.0

-- Disable submap creation
TRAJECTORY_BUILDER_2D.submaps.num_range_data = 999999

-- Only use existing map
POSE_GRAPH.optimize_every_n_nodes = 0

return options

​
Load Map

Launch with pre-built map: 
ros2 run cartographer_ros cartographer_node \
  -configuration_directory /path/to/config \
  -configuration_basename cartographer_localization.lua \
  -load_state_filename /home/user/maps/cartographer_state.pbstream
Or in launch file: 
arguments=[
    '-configuration_directory', cartographer_config_dir,
    '-configuration_basename', 'cartographer_localization.lua',
    '-load_state_filename', '/home/user/maps/cartographer_state.pbstream'
]

​
Advanced Features

​
Multi-Trajectory SLAM

For multi-robot or multi-session mapping: Start new trajectory: 
ros2 service call /cartographer/start_trajectory \
  cartographer_ros_msgs/srv/StartTrajectory \
  "{configuration_directory: '/path/to/config', configuration_basename: 'cartographer.lua'}"
Merge trajectories:
  • Automatically handled if robots share landmarks
  • Manual merge via services

​
3D SLAM

For 3D LiDAR (e.g., Velodyne): Config changes: 
MAP_BUILDER.use_trajectory_builder_2d = false  -- Disable 2D
MAP_BUILDER.use_trajectory_builder_3d = true   -- Enable 3D

TRAJECTORY_BUILDER_3D.num_accumulated_range_data = 1
TRAJECTORY_BUILDER_3D.min_range = 0.5
TRAJECTORY_BUILDER_3D.max_range = 60.0
TRAJECTORY_BUILDER_3D.submaps.high_resolution = 0.1
TRAJECTORY_BUILDER_3D.submaps.low_resolution = 0.5

​
Tuning Parameters

​
Scan Matching

-- Local scan matching (fast)
TRAJECTORY_BUILDER_2D.use_online_correlative_scan_matching = true

-- Ceres optimization weights
TRAJECTORY_BUILDER_2D.ceres_scan_matcher.translation_weight = 10
TRAJECTORY_BUILDER_2D.ceres_scan_matcher.rotation_weight = 40
Higher translation_weight: Trust odometry position more Higher rotation_weight: Trust odometry rotation more

​
Loop Closure

-- How often to attempt loop closure
POSE_GRAPH.optimize_every_n_nodes = 90

-- Minimum score for loop closure
POSE_GRAPH.constraint_builder.min_score = 0.55  # 0.0-1.0

-- Stricter score for global localization
POSE_GRAPH.constraint_builder.global_localization_min_score = 0.60
Lower min_score: More loop closures (risk of false positives) Higher min_score: Fewer loop closures (more conservative)

​
Motion Filter

-- Add scan only if robot moved this much
TRAJECTORY_BUILDER_2D.motion_filter.max_distance_meters = 0.2
TRAJECTORY_BUILDER_2D.motion_filter.max_angle_radians = math.rad(10)
TRAJECTORY_BUILDER_2D.motion_filter.max_time_seconds = 5.0

​
Comparison: Cartographer vs SLAM Toolbox

FeatureCartographerSLAM Toolbox
Ease of UseComplex (Lua config)Simple (YAML config)
2D SLAMExcellentExcellent
3D SLAMSupportedNo
Loop ClosureAdvancedGood
Real-TimeYes (but heavier)Yes
CPU UsageHighModerate
Memory UsageHighModerate
Multi-FloorExcellentLimited
Localization ModeYesYes
Map ExportPGM + PBStreamPGM + PoseGraph
ROS2 SupportGoodExcellent
Recommendation:
  • Start with SLAM Toolbox (simpler, works well for most cases)
  • Use Cartographer if:
    • Multi-floor building
    • Need 3D mapping
    • Have powerful computer
    • Complex environment with many loop closures

​
Troubleshooting

Debug:
  1. Check Lua config syntax: 
    # Test Lua file
lua config/cartographer.lua
  2. Check Cartographer node logs: 
    ros2 node logs /cartographer_node
  3. Verify topics remapped correctly: 
    ros2 topic list
# Should see /scan, /odom, /imu
Solutions:
  • Reduce scan frequency: 
    options.rangefinder_sampling_ratio = 0.5  -- Use 50% of scans
  • Increase motion filter thresholds (fewer scans added)
  • Disable online correlative scan matching (slower but less CPU)
  • Use async mode if available
Solutions:
  • Tune scan matching weights
  • Adjust loop closure parameters
  • Reduce mapping speed
  • Improve odometry calibration
  • Enable IMU fusion
Solutions:
  • Lower min_score threshold
  • Increase loop_search_maximum_distance
  • Drive slower (better scan matching)
  • Ensure sufficient feature overlap when revisiting

​
Next Steps

Navigation

Use Cartographer maps for autonomous navigation

Map Server

Load and serve maps for navigation

LiDAR Setup

Optimize LiDAR configuration for Cartographer

Advanced SLAM

Official Cartographer documentation

​
References

[1] Cartographer ROS: https://google-cartographer-ros.readthedocs.io/ [2] Cartographer Paper: https://research.google/pubs/pub45466/ [3] Cartographer Configuration: https://google-cartographer.readthedocs.io/en/latest/configuration.html