Simultaneous Localisation and Mapping (SLAM)

Introduction

SLAM is basically trying to map out an unknown environment while figuring out where it is in this unknown environment at the same time. We need to provide 2 streams of data to SLAM node.

1. Distance/Range Sensor

   • LiDAR (Laser Scanner) or a Depth Camera
   • Publishes to sensor_msgs/msg/LaserScan
   • This gives SLAM algorithm precise distance to walls, tables, etc

2. Odometry

   • Wheel Encoders + Inertial Measurement Unit (IMU)
   • It publishes nav_msgs/msg/Odometry
   • This tells SLAM roughly how far the wheels think the robot travelled

3. Unified Robot Description Format (URDF)

   • We need this so that we can generate the TF Tree which is being fed to the slam_toolbox
   • We must use a robot_state_publisher node to launch URDF file

Software Tools

There are some industry standard open-source packages:

1. slam_toolbox

   • modern default SLAM package for ROS 2
   • basically whatever we want to change, like odom and laser topic names we have are not default, we will just edit the YAML file

2. cartographer

   • Highly accurate but advanced to set up

3. nav2_map_server

   • utility tool used to actually save a finished map from terminal and load it back up later for navigation