icpslam

A basic SLAM system that employs 2D and 3D LIDAR measurements

TODO

• Class for handling 6 DOF poses, with time stamp, position, rotation and covariance.
• Handle robot odometry.
• Handle absolute robot pose from Gazebo.
• Estimate odometry using ICP on LIDAR measurements.
  • Cloud skipping for coping with sensors with high output frequency.
• Mapping
  • Implement a class to handle the current map with an Octree.
  • Localization on currently built map. (inspiration: BLAM!)
• Pose graph optimization
  • Abstract class for pose optimization.
  • Visualization with ROS markers.
  • Integrate g2o.
  • Integrate GTSAM.
  • Basic pose graph with ICP transforms. ("Smoothing")
  • Integrate wheel odometry.
• Technical improvements
  • Make odometry, localization and mapping nodes run concurrently.
  • Make everything thread safe.
• Wiki
  • Installation.
  • Parameters.
  • ROS nodes.
  • Examples of use.