/ur5e_simulation

ur5e_simulation

Primary LanguageC++

ur5e_simulation

ur5e_simulation

1. Requirements

System: Ubuntu 18.04, Ubuntu 20.04

ROS: Melodic, Noetic

Moveit

Gazebo

2. Other Dependencies

For some dependencies were missed for the project, please check the following instructions.

sudo apt-get install ros-melodic-moveit-visual-tools
sudo apt-get install ros-melodic-ros-controllers

Furthermore, one more gripper plugin was needed for the project.

cd <home>
git clone https://github.com/wangarcher/plugins.git
cd roboticsgroup_gazebo_plugins
mkdir build && cd build
cmake ..
make

PLEASE BE ADVISED!The following command would directly write a new .so lib into the system lib folder. It's effective, but a little bit dangerous.

sudo cp devel/lib/libroboticsgroup_gazebo_mimic_joint_plugin.so /usr/lib/x86_64-linux-gnu/<gazebo-9>or<gazebo-11>/plugins/

**NOTE:**There might be more dependencies needed due to vairous reasons. If you found any, please let me know.(archer7wang@outlook.com)

3. Usage

Like ordinary ros projects, the simluation needs a clean workspace.

mkdir -p ur5e_sim_ws/src
cd src
git clone https://github.com/wangarcher/ur5e_simulation.git
cd ..
catkin_make
source devel/setup.bash
1. Initialization

To initialize the ur5e_simluation

roslaunch sim_bringup with_moveit_new.launch
2. Planning

To run the demo visual-planning

roslaunch sim_general_planning x_arm_planning.launch
3. Control

To run the cartesian pid control

roslaunch sim_general_planning unity.launch

Others

About twelve_monkey

Find a reasonable manipulation path through tons of object poses.

roslaunch sim_general_planning twelve_monkeys.launch

The method is simple.

  1. Assume there is ellipse which focus on the start and end poses.
  2. Find k nearest object poses.
  3. Rank them according to the distance to the start and end poses.
  4. Done

Result

  • Bule point: object poses
  • Pink arrow: Manipulation pose
  • Green Path: Manipulation path result
About hex_love

An alternative choise

roslaunch sim_general_planning hex_love.launch

The method is simple, too.

K-means.

Result result