/trajectory_generation

ROS package that contains implementation of trajectory generation methods such as Model Predictive Control methods.

Primary LanguageC++BSD 3-Clause "New" or "Revised" LicenseBSD-3-Clause

trajectory_generation

ROS package that contains implementation of trajectory generation methods such as Model Predictive Control methods.

Dependencies

This is tested with Ubuntu 22.04, and ROS2 humble.

Setup

A sample setup.sh script is availble for convenience in the main directory of this repo.

Libraries

Core libraries are implementations of MPC with specific model with no dependence on ROS.

  • mpc_6dof.cpp implements MPC problem for a 3D discrete linear model with position and velocities as states, and acceleration as control inputs.

Nodes

mpc_ros

  • Uses mpc_6dof.cpp

Subscribers

  • px4_ros/in/odom : Message type nav_msgs::msg::Odometry. This is to get current state (e.g. position, velocity)
  • traj_predictor/in/ref_traj: Message type custom_trajectory_msgs::msg::StateTrajectory. This subscribes to the desired (reference) trajectory (position, velocity).

Publishers

  • mpc_tracker/out/path : Message type nav_msgs::msg::Path. To publish the poses of the solution state trajectory to visualize in RViz2
  • mpc_tracker/out/trajectory: Message type custom_trajectory_msgs::msg::StateTrajectory. Publishes the detailed trajectory (position, velocity, acceleration, ...)

config file

Run

ros2 launch trajectory_generation mpc.launch.py

Test

The core algorithm can be tested using test cases implemented in the test directory, no dependency on ROS.

  • test_mpc_6dof 
    ros2 run trajectory_generation test_mpc_6dof