This ROS2 Humble package contains a plugin for the Nav2 Controller Server that implements the Vector Pursuit path tracking algorithm. It leverages Screw Theory to achieve accurate path tracking and comes with active collision detection. The controller has a very low computational overhead and is very easy and simple to deploy. It tracks path orientation in a geometrically-meaningful way making it an ideal replacement for the Pure Pursuit Algorithm in scenarios where path following accuracy is vital. It consumes 15% of a single core on an ARM cortex-A72 CPU @ 1.8GHz and is designed to track paths at speeds upto 4.5m/s.

These are minimal, to-the-point instructions for experienced ROS2/Nav2 users. Beginners are recommended to read the Quickstart Tutorial for a simple 4-step example to try out the controller in a Gazebo simulation.

1. Install the package binaries.

   sudo apt-get install ros-humble-vector-pursuit-controller

2. Edit the controller_server parameters of the Nav2 stack to include the vector pursuit plugin along with its default configuration. Nav2's controller server supports multiple controller plugins at the same time and instructions for setting it up can be found in the official docs.

3. Build the workspace and source it, then run the nav2 stack/controller server.

These are additional features on top of the core Vector Pursuit algorithm that extend its functionality.

The following parameters tune the core path-tracking algorithm and are not needed by the additional features.

These parameters are used to tune and control the behaviour of

controller_server:
  ros__parameters:
    use_sim_time: True
    controller_frequency: 20.0
    min_x_velocity_threshold: 0.001
    min_y_velocity_threshold: 0.5
    min_theta_velocity_threshold: 0.001
    failure_tolerance: 0.3
    progress_checker_plugin: "progress_checker"
    goal_checker_plugins: ["general_goal_checker"]
    controller_plugins: ["FollowPath"]

    # Progress checker parameters
    progress_checker:
      plugin: "nav2_controller::SimpleProgressChecker"
      required_movement_radius: 0.25
      movement_time_allowance: 10.0

    # Goal checker parameters
    general_goal_checker:
      plugin: "nav2_controller::SimpleGoalChecker"
      xy_goal_tolerance: 0.25
      yaw_goal_tolerance: 0.25
      stateful: True
    
    FollowPath:
      plugin: "vector_pursuit_controller::VectorPursuitController"
      k: 5.0
      desired_linear_vel: 0.5
      min_turning_radius: 0.25
      lookahead_dist: 1.0
      min_lookahead_dist: 0.5
      max_lookahead_dist: 1.5
      lookahead_time: 1.5
      rotate_to_heading_angular_vel: 0.5
      transform_tolerance: 0.1
      use_velocity_scaled_lookahead_dist: false
      min_linear_velocity: 0.0
      min_approach_linear_velocity: 0.05
      approach_velocity_scaling_dist: 0.5
      max_allowed_time_to_collision_up_to_target: 1.0
      use_collision_detection: true
      use_cost_regulated_linear_velocity_scaling: true
      cost_scaling_dist: 0.5
      cost_scaling_gain: 1.0
      inflation_cost_scaling_factor: 3.0
      use_rotate_to_heading: true
      allow_reversing: false
      rotate_to_heading_min_angle: 0.5
      max_angular_accel: 3.0
      max_linear_accel: 2.0
      max_lateral_accel: 0.2
      max_robot_pose_search_dist: 10.0
      use_interpolation: true
      use_heading_from_path: false
      approach_velocity_scaling_dist: 1.0

We require a robot running the Nav2 stack to use the Vector Pursuit Controller. This example will utilise BCR Bot, a simulated, differential-drive robot with a sample Nav2 stack. An installation of ROS2 Humble is needed along with Git on an Ubuntu Machine.

1. Install the package binaries.

   sudo apt-get install ros-humble-vector-pursuit-controller

2. Install other ROS2 dependencies required for this tutorial.

   sudo apt-get install -y ros-humble-ros-gz-sim ros-humble-ros-gz-bridge ros-humble-ros-gz-interfaces ros-humble-bcr-bot ros-humble-navigation2 ros-humble-nav2-bringup  

3. Run the ROS2 demo launch file.

   ros2 launch vector_pursuit_controller bcr_bot_demo.launch.py

Vector pursuit is a geometric path-tracking method that uses the theory of screws. It is similar to other geometric methods in that a look-ahead distance is used to define a current goal point, and then geometry is used to determine the desired motion of the vehicle. On the other hand, it is different from current geometric path-tracking methods, such as follow-the-carrot or pure pursuit, which do not use the orientation at the look-ahead point. Proportional path tracking is a geometric method that does use the orientation at the look-ahead point. This method adds the current position error multiplied by some gain to the current orientation error multiplied by some gain, and therefore becomes geometrically meaningless since terms with different units are added. Vector pursuit uses both the location and orientation of the look-ahead point while remaining geometrically meaningful. It first calculates two instantaneous screws for translation and rotation, and then combines them to form a single screw representing the required motion. Then it calculates a desired turning radius from the resultant screw.

The centerlines of the instantaneous screws are constrained to lie on the vehicle's y-axis to accomodate one of the non-holonomic contraints. The screw for correcting the translational error, $_t is chosen as the center of a circle which intersects the origins of both the vehicle coordinate system and the look-ahead coordinate system and is tangent to the vehicle's current orientation. The screw for correcting the rotational error, $_r is chosen to be centered at the vehicle frame origin so that no translation is associated with it.

To generate an HTML webpage to view code coverage use:

genhtml code_coverage_report.info --output-directory ~/vector_pursuit_code_coverage_report

We acknowledge the contributions of:

1. The author of Vector Pursuit Path Tracking for Autonomous Ground Vehicles, Jeffrey S. Wit.
2. The Nav2 Regulated Pure Pursuit Controller project.