Skip to content

README_f1tenth_amcl.md

Latest commit

 

History

History
62 lines (55 loc) · 2.65 KB

README_f1tenth_amcl.md

File metadata and controls

62 lines (55 loc) · 2.65 KB

F1Tenth Localization Algorithm: Particle Filter

This exercise is based on CMU 16663 - F1Tenth Course :: Lab 5.

I. Learning Goals

Here, we are getting into localization, specifically, we are going to learn about and use particle filter.

Please watch UPenn's F1Tenth lecture on Localization using Particle Filters and review CMU's F1Tenth lecture slide on Particle Filters and Particle Filter Analysis.

II. Particle Filter

Once you have a comprehensive understanding of localization with particle filters, run particle_filter on the car using the map you made with SLAM inside the simulator.

  1. Clone the range_libc repo: 
    git clone https://github.com/f1tenth/range_libc.git
  2. Install with the following commands: 
    cd range_libc/pywrapper
    sudo WITH_CUDA=ON python setup.py install
  3. Clone the particle_filter package 
    cd /home/{username}/{workspace}/src
    git clone https://github.com/f1tenth/particle_filter.git
  4. Install dependencies 
    rosdep install -r --from-paths src --ignore-src --rosdistro foxy -y
  5. Compile your workspace again and source it 
    cd /home/{username}/{workspace} && colcon build
    source install/setup.bash
  6. Launching particle_filter
    • Launch teleop in one window
    • Launch particle_filter in another window 
      ros2 launch particle_filter localize_launch.py
  7. Visualization
    • Launch rviz2
    • Add /map by topic
    • In the settings for /map, change durability policy under topic to transient local.
    • To show the current localization, add /pf/viz/inferred_pose topic.
    • Optionally, you can add /pf/viz/particles to see the particles.
  8. Check the update frequency Check the publishing frequency on /pf/viz/inferred_pose. It should be at least 30 Hz.
  9. Changing the map in use
    • Put the map files (.png image and .yaml file) in particle_filter/maps.
    • Change particle_filter/config/localize.yaml to reflect the map you want to use.
  10. Use the "2D Pose Estimate" to set the initial position for the particle filter, as shown below in the figure: