barrett-ros-pkg

ROS package for the Barrett WAM and related products.

• Overview
• Pre-Requisites
• Compiling the package
• Running wam_node
• Running wam_demos
• Running perception_palm
  • Set up cameras
  • Calibration
  • Running the demo
  • Accessing the sensors
  • LED
  • Laser
  • IR Range finder
  • Camera
  • Networking
  • Troubleshooting
• Running barrett_hand_node
• Example of running the services (Tested on ROS Melodic and Indigo)

Overview

This is Barrett Technology's ROS repository that wraps Libbarrett's functionalities and includes a ROS driver for Barrett's Perception Palm. The Perception Palm includes a LED, a Laser, two cameras and an IR Range finder. The driver for the Perception palm wraps the open source C/C++ library for Microchip's USB-to-SPI protocol coverter. Libbarrett is a real-time controls library written in C++ that runs Barrett Technology's products, including the WAM Arm and the BH8-282 BarrettHand.

• The wam_node stack is designed to be run on a WAM PC-104 or external control PC, and can work with the WAM with any combination of the wrist, BarrettHand and the Force/Torque sensor attached.
• The barrett_hand_node stack is designed to run on an external control PC and can work with the BarrettHand standalone connected via the CAN bus.
• The perception_palm stack is designed to run on an external control PC.
• The wam_msgs, wam_srvs and wam_teleop stacks are designed as the interface to communicate with the functionality exposed by the wam_node.

Pre-Requisites

On Ubuntu 14.04:

1. An installed version of Libbarrett 1.3.0

2. ROS Indigo

3. Install libudev and wstool.

sudo apt-get update
sudo apt-get install libudev-dev
sudo apt-get install python-wstool

4. Install the camera driver:

sudo apt-get install ros-indigo-camera-umd

On Ubuntu 18.04:

1. An installed version of Libbarrett 2.0.0

2. ROS Melodic

3. Install libudev and wstool.

sudo apt-get update
sudo apt-get install libudev-dev
sudo apt-get install python-wstool

4. Install the camera driver:

sudo apt-get install ros-melodic-camera-umd

On Ubuntu 20.04:

1. An installed version of Libbarrett 3.0.0

2. ROS Noetic

3. Install libudev and the camera driver.

sudo apt update
sudo apt install libudev-dev ros-noetic-usb-cam

Compiling the package

Create a new Catkin Workspace if not already done:

mkdir -p ~/catkin_ws/src
cd ~/catkin_ws/
catkin_make

Clone the git repository an run the build script:

cd ~/catkin_ws/src
git clone https://git.barrett.com/software/barrett-ros-pkg.git
cd barrett-ros-pkg
git checkout devel
sudo -s
./build.sh
exit

Source the package before running (does not persist after reboot):

source ~/catkin_ws/devel/setup.bash

OR add it to bashrc (persists after reboot):

echo 'source ~/catkin_ws/devel/setup.bash' >> ~/.bashrc
source ~/.bashrc

Running wam_node

Launch the wam_node.launch file, with the WAM Arm connected:

roslaunch wam_node wam_node.launch

Running wam_demos

To Cycle a 7DOF WAM and BarrettHand 10 times from a shell script:

sh cmds-7dof-cycle.sh 10

To Teach:

rosrun wam_demos teach -t <record_type> -n <bag_name>

The -t <record_type> field allows you to choose how to record the trajectories Possible values are:

• -t jp: Record using joint positions
• -t jv: Record using joint positions

To Play:

rosrun wam_demos play <bag_name>

Running perception_palm

Set up cameras:

1. Install the camera utilities

sudo apt install v4l-utils guvcview

2. Connect the Perception Palm to the PC using a USB cable

3. Determine the device names for the Perception Palm cameras.

List the video devices with:

v4l2-ctl --list-devices

To determine which devices are correct, you can use a program such as guvcview:

guvcview -d /dev/video0

Note which device corresponds to which camera. Some of the /dev/video devices may actually correspond to a camera's metadata (like image format) instead of representing a video stream. For detailed information on /dev/video0 (for example), use:

v4l2-ctl --device=/dev/video0 --all

If the "Device Caps" section advertises "Video Capture", then this device has a valid video stream.

4. Edit the launch file to confirm camera setup parameters.

gedit ~/catkin_ws/src/barrett-ros-pkg/perception_palm/launch/perception_palm.launch

Ensure that the launch file targets the correct devices. If you need to change the default devices, edit the lines in the launch file that look like this:

<param name="device" type="string" value="/dev/video0" />

Notes

A single camera can be used at a maximum resolution of 1600 x 1200. Two cameras are limited to a maximum resolution of 320 x 240 due to how much USB bandwidth is (incorrectly) requested/estimated for each video stream. For information on alternative camera resolutions, formats, and frame rates, use this command:

v4l2-ctl --list-formats-ext

The camera with the red filter is physically rotated 180 degrees with respect to the other camera. You may want to flip the video stream from this camera.

Calibration

IR Range finder

The IR range finder needs to be calibrated for the first time before using it. It will work without calibration but it might not be accurate. Follow the instructions below to calibrate it. This step can be skipped if you do not want to use the IR or if the IR range finder is already calibrated.

cd ~/catkin_ws/src/barrett-ros-pkg/perception_palm/include/MCP2210-Library
make
sudo ./dist/Debug/GNU-Linux-x86/ir_calibrate

Follow the onscreen instructions.

Cameras

Please refer to the ROS Stereo/Monocular calibration package. This step is completely optional. The cameras would work even if this step is skipped.

Running the demo

1. (every time you plug in the Perception Palm or reboot the computer). For one camera

sudo rmmod uvcvideo
sudo modprobe uvcvideo

or two cameras

sudo rmmod uvcvideo
sudo modprobe uvcvideo quirks=128

2. Become the root user to access the drivers and run the demo. For one camera:

sudo -s
source /home/robot/catkin_ws/devel/setup.bash
roslaunch perception_palm perception_palm.launch

For two cameras:

sudo -s
source ~/catkin_ws/devel/setup.bash
roslaunch perception_palm perception_palm.launch mono_camera:=false

3. To quit, press Ctrl-C. Then type exit to return to a regular terminal.

Troubleshooting

If the camera node fails to start in step 2, make sure the configuration you chose in step 1 matches the configuration in the launch file. See the "Set up cameras" section for details.

Accessing the sensors

While the demo is running you can access the sensors from a separate terminal.

LED

The LED can be turned off and on by calling the service barrett/palm/set_led_on.

rosservice call barrett/palm/set_led_on ['True']
rosservice call barrett/palm/set_led_on ['False']

Laser

The Laser can be turned off and on by calling the service barrett/palm/set_laser_on.

rosservice call barrett/palm/set_laser_on ['True']
rosservice call barrett/palm/set_laser_on ['False']

IR Range finder

The IR range finder publishes the range information as a sensor_msgs/Range message, at 1Hz to the topic barrett/palm/ir/range

rostopic echo barrett/palm/ir/range

Camera

By default, the two camera feeds are published at the rate of 30 fps with a resolution of 320x240 px. They are published to the topics, barrett/palm/left/image_raw and barrett/palm/right/image_Raw.
While using the monocular camera mode, the images are published at the rate of 30 fps with a resolution of 1600 x 1200 to the topic barrett/palm/image_raw.

Networking

This is required if you use run the ROS nodes across multiple computers. The below instructions are to setup the perception palm on the XWAM and to control it from a remote host.

In the host computer, open a new terminal and ssh into the XWAM

ssh summit@*xwam's ip address*

Enter the XWAM's password and open the hosts file

sudo vi /etc/hosts

Add the IP address of the remote host machine to the list and name it

*remote host's ip address* *remote host's name*

Save and exit it. Set the XWAM to be the ROS MASTER

export ROS_MASTER_URI=http://*XWAM's ip address*:11311

Open a new terminal and edit the host file in the host machine as above

sudo vi /etc/hosts

In the host machine, name the IP address of the xwam

*xwam's ip address* summit

Save and exit it. Update the ROS_MASTER as the XWAM in the host machine

export ROS_MASTER_URI=http://*XWAM's ip address*:11311

P.S: The ROS_MASTER_URI variable would be active as long as the terminal is open. If the terminal is closed and restarted, the ROS_MASTER_URI needs to be updated. For more information on ROS network configuration refer to ROS documentation.

Troubleshooting

Trying to restart the perception_palm package multiple times might fail with an error "Error setting SPI Parameters". This can be solved by unplugging and plugging back the USB to the port.

Running barrett_hand_node

Launch the barrett_hand_node.launch file, with the BarrettHand connected via the CAN bus:

roslaunch barrett_hand_node barrett_hand_node.launch

In a separate terminal:

rosservice call /bhand/close_grasp
rosservice call /bhand/open_grasp
rosservice call /bhand/grasp_pos 1.57
rosservice call /bhand/finger_pos "[0.5, 1, 1.5]"
rosservice call /bhand/initialize
rosservice call /bhand/spread_pos 1.57
rosservice call /bhand/open_spread

rostopic echo /bhand/finger_tip_states
rostopic echo /bhand/joint_states
rostopic echo /bhand/tactile_states

Examples of running the WAM services (Tested on ROS Melodic and Indigo)

Move WAM Joints:

rosservice call /wam/joint_move "joints:
- 0.0
- 0.0
- 0.0
- 0.0"

Move WAM to Tool Pose:

rosservice call /wam/pose_move "pose:
  position:
    x: 0.048
    y: 0.0
    z: 0.618
  orientation:
    x: -0.0190
    y: 0.9022
    z: -0.2516
    w: -0.3498"

Move WAM Home:

rosservice call /wam/go_home

Hold Joint Positions:

rosservice call /wam/hold_joint_pos "hold: true"

Unhold Joint Positions:

rosservice call /wam/hold_joint_pos "hold: false"