Kinodynamic Motion Planning in Dynamic Environment for robot soccer

click here to view full video. Or here for more videos.

Detailed Description

Read the thesis or paper

You'll need :

• ROS
• gcc >= 7 (for C++17)
• Cython
• google benchmark

Cloning

do recursive cloning :

git clone --recursive https://github.com/alifahrri/robosoccer_motion_planning.git

Building

standard ROS build, but it is recommended to use only single thread since it will take huge amount of RAM.

catkin_make -j1

Running Unit Tests

run unit test setting from catkin

catkin_make -j1 run_tests

Run Benchmark

use google benchmark settings

rosrun rrtstar rrtstar_benchmark

custom settings (report min, max, mean also prints individual test results)

rosrun rrtstar rrtstar_custom_benchmark

Performance

tested on intel i3 2.0 GHz

computation time	solution cost

no direct sampling	direct sampling with prob = 0.1

Nodes

• the core of the motion planning algorithm robosoccer_motion_planning/src/robosoccer_motion_planning_node
• trajectory tracking node for robot soccer robosoccer_trajectory_tracking/scripts/trajectory_cytracker.py
• control robots with waypoints robosoccer_teleop/scripts/robosoccer_teleop_node.py
• example of trajectory generator as server node robosoccer_motion_planning/src/trajectory_generator_node
• example of trajectory generator client node robosoccer_motion_planning/scripts/trajectory_demo.py

Launcher

• run all necessary nodes (core, trajectory tracking, and teleop) : robosoccer_bringup.launch
• record necessary messages with rosbag : recorder.launch. Launch it after bringup.
• play recorded bag files (this will shows visualisation and plotter) : player.launch file:=your_file_full_path.bag