TIAGo hardware
From Robin
This page contains information about the different attachments of TIAGo, like the endoscopic camera, NVIDIA Jetson, LEAP motion, and the tablet.
If you wish to use any of these add-ons, contact robin-engineer@ifi.uio.no.
Contents |
Endoscopic camera
The endoscope is a small camera that can be attached to TIAGo's end-effector, in order to visualize the view from the perspective of the end-effector. To set up the endoscope and run it, follow these steps:
1. Follow the instructions in the TIAGo handbook, page 301-306. Skip the part about visualizing the camera output, and instead, follow the next steps.
2. In another terminal on the development computer, enter the container:
$ singularity shell --nv /projects/robin/programs/sing-con/ros-melodic-pal/ros-melodic-pal.sif
3. Source ROS and the workspace:
$ source /opt/ros/melodic/setup.bash $ source /tiago_public_ws/devel/setup.bash
4. Remember to have the correct environment variables set:
$ export ROS_MASTER_URI=http://tiago-135c:11311 $ export ROS_IP=vixen-nb.uio.no
5. Then run the following command to run Rviz and bring up the robot:
$ rosrun rviz rviz -d `rospack find tiago_bringup`/config/tiago.rviz
6. To visualize the image, click on the buttons "Add" and then "By topic".
7. Find the topic called /end_effector_camera. Click on /image_raw and then on "Image", as shown in the picture below.
NVIDIA Jetson
Tiago has an NVIDIA Jetson TX2 device that can be attached to TIago for extra processing power for heavier jobs. You can read more about it in the Tiago handbook from page 281.
The handbook contains a tutorial for how to perform object detection with Tiago and the NVIDIA Jetson. I have tried it out and it didn't work as expected, because it looks like one of the processes keep dying. One of the things you have to do in the tutorial is run an image_view window where you display the output from Tiago's camera as well as the object detections published on the /jetson_detections topic (this is step 2). This window never pops up and we get an error that the process has died and this is where the tutorial fails in practice.
- image_view works fine in other instances, so it is unknown what the issue here is - therefore it is possible that the NVIDIA Jetson works fine for other problems and that it is just the tutorial that is troublesome.
LEAP motion
Tiago also has a leap motion device that allows you to control Tiago's manipulator with hand gestures that you perform yourself using the device. The Tiago handbook contains information about this from page 243.
The handbook has a detailed tutorial for this device. The issue with this tutorial is that you need superuser rights to run the code, something we don't have on the vixen development computer in the mocap lab or within the container. Therefore, we get a related error when trying to follow the tutorial.
- The leap motion device needs to be connected to the development computer and therefore you need access to the USB port on the computer, but this was not the source for this issue.
- I tried running the container as fakeroot (https://apptainer.org/docs/user/1.0/fakeroot.html) to run the container with almost the same rights as an administrative user. Within the container you can run things as superuser then, but this didn't solve the issue as I still get the same error.
- It is recomended to use the vixen computer with Tiago since it is on the same network as Tiago and all the ports are open between the computer and robot, but maybe it is possible to use the leap motion on a private development computer instead where you have administrative rights.
Tablet
Tiago also has a tablet with a touch screen that can be used for human-robot interactions.
