This repository contains simulations of the Universal Robots UR3 robot created for the ERC competition. You can use docker with all requirements installed (Using Docker section) or try to install them natively (Install on the host system section).
The simulation uses the ROS-Industrial Universal Robot repository and roboticsgroup_gazebo_plugins. The following items have been added to the simulation: robot cell, table surface with marked areas, four button modules with ID 1 - 4 (in a pressed down state, not movable), inspection window along with its cover, the IMU module and gripper with camera. The files necessary for motion planning have also been modified to take into account the added elements. All changed and added files are located in the UR3_sim repository. You can run the simulation with an example cell or only with a robot, gripper and camera.
The gripper used was taken from the Printables website.
A block has been added to the gripper in the place where the plugs are. This will avoid collisions with plugs.
NOTE
The commands in this section should be executed as the root user, unless you have configured docker to be managable as a non-root user.
Make sure the Docker Engine is installed and the docker service is running:
systemctl start docker
Build the docker image by executing:
./build_erc.sh
Create container: If you are running the system with no dedicated GPU (not recommended, Gazebo may not run or work slowly without the GPU), execute:
bash run_ur3_docker.bash
To use an integrated AMD/Intel Graphics card, run:
bash run_ur3_docker_AMD_Intel_Graphics.bash
To use an Nvidia card, you need to previously install proprietary drivers and Nvidia Container Toolkit (https://github.com/NVIDIA/nvidia-docker). Next execute command:
bash run_ur3_docker_Nvidia.bash
Then, you can start container:
docker start ur3_simulation_erc
Then, you can run command in running container:
docker exec -it ur3_simulation_erc /bin/bash
The simulation is mainly developed and tested on Ubuntu 20.04 Focal Fossa with ROS Noetic Ninjemys, so it is a recommended setup.
For the simulation to work properly, you must install dependencies and download repository by running the following commands:
source /opt/ros/noetic/setup.bash
sudo apt-get update && apt-get upgrade -y && apt-get install -y lsb-core g++
sudo apt-get install git
rosdep init && rosdep update
sudo apt install ros-noetic-moveit -y
sudo apt install ros-noetic-ros-controllers* -y
mkdir -p /catkin_ws/src
cd /catkin_ws/src
git clone -b kinetic-devel https://github.com/ros-industrial/universal_robot.git
sudo rm -r universal_robot/ur_msgs
git clone https://github.com/roboticsgroup/roboticsgroup_gazebo_plugins
git clone https://github.com/Michal-Bidzinski/UR3_sim.git
cd /catkin_ws
catkin_make
source devel/setup.bash
To run UR3 simulation in Gazebo with MoveIt! and RVzi GUI, including an example cell:
$ roslaunch ur3_sim simulation.launch
To run UR3 simulation in Gazebo with MoveIt! and RVzi GUI, containing only robot with gripper and camera:
$ roslaunch ur3_sim real_station.launch
To control the arm can by used MoveIt!. Planning robot movement can be performed using for example the Move Group Interface. Tutorials written in Python and C++ can be a hint (note the robot group name and number of joints).
The gripper is controlled by publishing appropriate commands on topic /gripper command. Message type: std msgs/String. To control the gripper, send the one of following commands:
- open
- semi_open (for catching the IMU box)
- semi_close (for catching the lid of the box)
- close
The simulation includes a camera placed on a gripper that detects aruco tags and other items. The following topics are published:
- /camera_image/camera_info
- /camera_image/image_raw
- /camera_image/image_raw/compressed
- /camera_image/image_raw/compressed/parameter_descriptions
- /camera_image/image_raw/compressed/parameter_updates
- /camera_image/image_raw/compressedDepth
- /camera_image/image_raw/compressedDepth/parameter_descriptions
- /camera_image/image_raw/compressedDepth/parameter_updates
- /camera_image/image_raw/theora
- /camera_image/image_raw/theora/parameter_descriptions
- /camera_image/image_raw/theora/parameter_updates
Although the topics with the word Depth in the name are published by defalut ros node, these cameras do not capture any depth data.