ros_libgp

A ros-wrapper for the libgp; a C++ library for Gaussian process regression. the goal of this wrapper/package is to have a ready-to-run ros package that uses libgp. with this package you can subscribe for a ros topic to get the data to train the GP, then you can predict the GPP output at differetn inputs, and finally you can publish the prediction results to another ros topic.

Dependencies

my libgp-fork: this fork of libgp has all the header files updates to reflect the new location of the libgp in the ros package
eigen3

Install

Create catkin workspace

$ mkdir -p gp_ws/src
$ cd gp_ws/src

Clone ros_libgp package using the --recurse-submodules option, this option will clone both the ros_libgp and my libgp-fork

$ git clone --recurse-submodules https://github.com/mahmoud-a-ali/ros_libgp.git

Compile workspace (either using catkin build or catkin_make)

# change directory to workspace "gp_ws"
$ cd ..   
# compile
$ catkin build    
# source the workspace
$ source devel/setup.bash

Run gp_node

# in one terminal run ros master
$ roscore
# in another terminal run the gp_node
$ rosrun ros_libgp gp_node