Skip to content

Model Predictive Control (MPC) based waypoint tracking controller (ROS navigation stack local planner plugin)

License

Notifications You must be signed in to change notification settings

balamuruganky/mpc_waypoint_tracking_controller

Repository files navigation

MPC Waypoint Tracking Controller (Local Planner for ROS Navigation)

Abstract

To move the robot on the predefined path accurately based on the plan shall be published by Global planner. Non linear model predictive controller shall help to calculate the cmd_vel based on the desired trajectory. This local planner shall be used along with WaypointsGlobalPlanner.

mpc_demo

Dependencies

- CppAD (included in this build)
- Ipopt (included in this build)
- Eigen (sudo apt install libeigen3-dev)

How to compile (Assumed that the catkin workspace setup already)

- cd catkin_ws/src
- git clone https://github.com/balamuruganky/mpc_waypoint_tracking_controller
- cd ..
- catkin_make

Configuration

Please refer MoveBase Parameters

Caveats

- Robot Position and orientation must be towards initial waypoint position and orientation.
- Waypoints must be dense enough to calculate the cmd_vel.

Demo

Please clone bbblue_bot to run demo of mpc_waypoint_tracking_controller along with waypoints_global_planner

How to run the demo

Rviz

roslaunch bbblue_navigation navigate_waypoints_slam.launch

Global Path Plan

roslaunch waypoints_global_planner desired_path_provider.launch

Change the Global plan

Modify the points based on your need in this file and relaunch the above global path plan launch file.

Please note that, the robot current pose must be the starting pose of the global path plan.

Credits

- https://github.com/Geonhee-LEE/mpc_ros
- https://github.com/Hypha-ROS/hypharos_minicar

TODO

- Test the controller with different desired trajectories
- Remove the dependency of robot current pose must be the start pose of the global plan.

Releases

No releases published

Packages

No packages published

Languages