PIPER (Probabilistic Inference based Platform for Essential problems in Robotics) is a modular package that provides support for algorithms that use probabilistic inference on factor graphs to solve various robotics problems. Each module can be independently installed and implements an easy ROS interface for that algorithm to run on any simulated or real robots. Currently PIPER supports the following algorithms:

• GPMP2 - Gaussian Process Motion Planner 2
• STEAP - Simultaneous Trajectory Estimation and Planning
• more coming soon ...

PIPER is being developed by Mustafa Mukadam at the Georgia Tech Robot Learning Lab. See documentation for information on usage and development.

• Prerequisites
• Compilation and Installation
• Questions and Bug reporting
• Citing
• License

• Install ROS. We have tested ROS indigo on Ubuntu 14.04.
• Install GPMP2 C++ library.

• Initialize a catkin workspace (if you are using an existing catkin workspace this step is not needed)

  mkdir -p ~/piper_ws/src
  cd ~/piper_ws/src
  catkin_init_workspace

  Before running setup the environment variables

  source ~/piper_ws/devel/setup.bash

• Clone this repository in ~/piper_ws/src

  git clone https://github.com/gtrll/piper.git

• To compile only the piperbase library, in the catkin workspace directory do

  catkin_make

• Otherwise, to install some module, for example, X: first make sure to install dependencies for X, besides the prerequisites for piperbase, then do

  catkin_make -build_flag_X

• Similarly, to install multiple modules, for example, X and Y: install all their dependencies and then use their appropriate flags together

  catkin_make -build_flag_X -build_flag_Y

  See table below for currently supported modules, their dependencies and build flags

  Module	Other Dependencies	Build Flag
  GPMP2	None	-DBUILD_GPMP2_INTERFACE:OPTION=ON
  STEAP	None	-DBUILD_STEAP_INTERFACE:OPTION=ON
  ALL	All from above	-DBUILD_ALL_INTERFACE:OPTION=ON

Please use Github issue tracker to report bugs. For other questions please contact Mustafa Mukadam.

If you use PIPER in an academic context, please cite any module/algorithm specific publications you use, and cite the following:

@article{mukadam2017piper,
  title={{PIPER}},
  author={Mukadam, Mustafa},
  journal={[Online] Available at \url{https://github.com/gtrll/piper}},
  year={2017}
}

PIPER is released under the BSD license. See LICENSE file in this directory.