Skip to main content

🗺️ SLAM (Simultaneous Localization and Mapping)

SLAM is the core that allows your robot to build a map of an unknown environment while keeping track of its own position—all in real time.
This is the foundation for autonomous navigation, exploration, and manipulation.

🤔 What is SLAM?​

SLAM stands for Simultaneous Localization and Mapping.
It lets your robot create a map and know where it is on that map—using only its own sensors.

  • Localization: "Where am I right now?"
  • Mapping: "What does my world look like?"
  • SLAM: "Where am I, and what does my world look like—at the same time?"

🧠 How SLAM Works (in Your Project)​

  • Your robot uses LiDAR, depth camera, and optionally IMU/wheel odometry for input.
  • The SLAM node (Cartographer or RTAB-Map) creates a map and constantly estimates robot pose.
  • The map and pose are published to other nodes (like Navigation).

⚙️ SLAM in This Project​

You can choose between two industry-standard SLAM implementations:

Best for: Fast, lightweight, real-time 2D mapping
Uses: LiDAR (YDLIDAR T-mini Pro) for map building

# Launch robot base and LiDAR
ros2 launch limo_bringup limo_start.launch.py

# Start Cartographer SLAM
ros2 launch limo_bringup cartographer.launch.py

Visualize in RViz:

rviz2

🗺️ Building and Saving a Map​

  1. Move the robot slowly throughout the environment (use teleop or remote/app)
  2. When done, save the map:
ros2 run nav2_map_server map_saver_cli -f my_map

Map will be saved as my_map.yaml and my_map.pgm in the current directory.

Map Files
  • .yaml: Map metadata (resolution, origin, thresholds)
  • .pgm: Actual map image (grayscale occupancy grid)

🎯 Localization Strategies​

After a map is built, you need localization to allow the robot to "find itself" on the map at startup and during navigation.

AMCL (Adaptive Monte Carlo Localization)​

Used after Cartographer mapping. Tracks the robot's position on a 2D map using LiDAR and motion data.

ros2 launch limo_bringup limo_nav2.launch.py
# Or for Ackermann mode:
ros2 launch limo_bringup limo_nav2_ackmann.launch.py

RTAB-Map Localization​

Launch rtabmap in localization mode:

ros2 launch limo_bringup limo_rtab_slam.launch.py localization:=true
Initial Pose Setup

In RViz, always set the robot's initial pose to align the laser scan with the map!
Use the "2D Pose Estimate" tool in RViz after startup.

🛠️ SLAM Configuration​

Key Parameters for Cartographer​

ParameterDescriptionTypical Value
map_resolutionMap cell size (meters)0.05
tracking_frameRobot base framebase_link
published_frameMap framemap
use_odometryUse wheel odometrytrue

Key Parameters for RTAB-Map​

ParameterDescriptionTypical Value
frame_idRobot base framebase_link
map_frame_idMap framemap
database_pathDatabase location~/.ros/rtabmap.db
localizationLocalization modefalse (mapping)

💡 Best Practices & Troubleshooting​

🐌 Move Slowly During Mapping
  • Move at 0.3 m/s or slower for best accuracy
  • Avoid sudden turns or rapid acceleration
  • Give the algorithm time to process sensor data
🔄 Loop Closure
  • Don't close loops too quickly—give the algorithm time to recognize previously seen areas
  • Ensure good lighting for camera-based SLAM
  • LiDAR-based SLAM is more robust to lighting changes
💾 Data Management
  • Backup your maps and SLAM database after every successful session!
  • Use descriptive filenames with dates: office_2025-08-07.yaml
  • Test map quality before relying on it for navigation
⚠️ Common Issues
  • Laser scan and map don't align: Manually set initial pose in RViz
  • Map quality poor: Check sensor mounting, move slower, ensure good lighting
  • Localization fails: Verify map files exist and are in correct format

🔧 Advanced SLAM Features​

Multi-Session Mapping​

# Continue mapping with existing RTAB-Map database
ros2 launch limo_bringup limo_rtab_slam.launch.py 
# Database automatically loads previous session

Online Loop Closure​

  • Cartographer: Automatic loop closure optimization
  • RTAB-Map: Real-time loop closure with visual and geometric validation

Map Updates​

  • Static environments: Maps remain stable over time
  • Dynamic environments: RTAB-Map can handle moving objects

📊 Performance Comparison​

FeatureCartographerRTAB-Map
Speed⚡ Very Fast🐌 Moderate
Map Quality📏 Good 2D🎯 Excellent 3D
Memory Usage💾 Low💾 High
Robustness🛡️ Very Good🛡️ Excellent
Setup Complexity🔧 Simple🔧 Moderate

📚 Learn More​

🎯 Next Steps​

Ready to Navigate?

Once you have a good map, you're ready for autonomous navigation!

on August 8, 2025 by KRISH RAY