Nonlinear Model Predictive Controller

About

This repository contains an MPC Trajectory Tracker based on the mpc_ros library developed by Geonhee-LEE et. al. The MPC itself is written over the IPOPT nonlinear solver and integrated with ROS (Robot Operating System). The model used for trajectory tracking within the MPC is the unicycle model, but there are plans to expand this out to double / triple integrators as well, and even to aerial vehicles. These features will come out when I either need them for a project or when I get the free time to develop them.

Installation

NOTE: Tested in Ubuntu 16.04/18.04/20.04 in ROS Kinetic/Melodic/Noetic.

1. Install ROS

2. Install IPOPT. So far only tested on IPOPT version 3.14.4, no guarantees that it works on any other version.

3. Clone repository into a catkin_ws and compile with either catkin build or catkin_make

   • Note: The following line was added to the CMakeLists.txt in case the user doesn't want to add /usr/local/lib to LD_LIBRARY_PATH:

   link_directories(/usr/local/lib)
   

   If you're IPOPT install is located elsewhere, I'd recommend changing that directory to wherever your IPOPT libraries are if you are having problems linking.

Running

You can run the MPC by launching the following launch file:

roslaunch mpc_trajectory_tracking jackal_mpc_track.launch

All the MPC needs is a reference trajectory, then it will publish a Twist message to the cmd_vel topic.

Note: You can change the default odometry topic via the odomTopic argument in the launch file.

You can send a trajectory to one of two topics:

1. /reference_trajectory: Will reset the trajectory timer and track the trajectory starting at time=0s.

2. /reference_trajectory_no_reset: Same as reference_trajectory but will not reset the trajectory timer. This is useful for appending to a trajectory currently being tracked.

Maintenance

I am actively adding features / fixing bugs in this repository. With that said, if you would like to use this repository and are having any troubles, please feel free to submit an issue and I will most likely see it within a few days.