1. Clone this repository into a directory

git clone [email protected]:Hartvi/jogramop-planners.git

2. Clone PQP into the same directory

• because in CMakeLists.txt there is: add_subdirectory(../PQP PQP_build)
• or move it where you like and change the CMakeLists.txt as well

git clone [email protected]:GammaUNC/PQP.git

3. Install the following libraries:

• urdfdom - sudo apt-get install liburdfdom-dev and liburdf-dev
• KDL - sudo apt-get install libkdl-parser-dev and liborocos-kdl-dev
• Eigen https://gitlab.com/libeigen/eigen/ / sudo apt install libeigen3-dev
  • The version used is 3.3.7-2
• FLANN https://github.com/flann-lib/flann / sudo apt install libflann-dev
• LZ4 https://github.com/lz4/lz4 / sudo apt install liblz4-dev

4. Clone https://github.com/mrudorfer/jogramop-framework into this directory

cd jogramop-planners
git clone [email protected]:mrudorfer/jogramop-framework.git

4. Then you can build it

mkdir build
cd build
cmake ..
make

5. Run planners using one of the P{i}_{planner}.sh scripts. There you can set parameters as you need.
6. Run ./show_options.sh to show all parameters options and their meanings

The code was tested on Ubuntu 20.04, with C++17.

The planner outputs three files (with a prefix specified by -target_prefix PREFIX

• PREFIX.try - this the trajectory, a list of configurations on each line
• PREFIX.vis - the file containing the Frame of the obstacles and sobot segments to visualize the path in 3D
• PREFIX.txt - this file contains measurements. Success, distance to target, time, number of iterations, tree size. e.g. { "sr":1,"dtg":43.0474,"time":0.130713,"iters":118,"treesize": 120}

• NOTE: At the moment only available on the master branch (main branch coming soon). Run git checkout master to perform experiments.

For statistical evaluation of the planners, it's necessary to run the planners several time (on each scene or for each planner settings). The basic testing can be achieved in this way:

python3 runExperiments.py

The script will create all-cmds.sh, which contains command lines for each planner/scenario. The planners can be run either sequentially (i.e., running sh all-cmds.sh), or by parallel

parallel < all-cmds.sh

After experiments finish, the resuls are in the results/ directory. Basic processing (drawing success rate curves into png) can be realized by>

python3 makeSucessRate