Hello,
I’m having trouble starting the practical test for my Stretch-robot 3. Can someone help me or send me videos or anything else that might be helpful?
Hi Mallak,
Thanks for reaching out! Happy to help.
To better understand what’s going on, could you please share a bit more detail about where you’re getting stuck?
In the meantime, we do have a number of demonstrations and walkthroughs available on our website. I’d recommend taking a look at our official documentation here:
The Getting Started section is a good place to begin and may help clarify some of the initial setup and testing steps.
Best,
Jason
I have struggle when i want to do Language-Directed Pick and Place demo, i followed all steps but still there is no good results, also i’m little confused about how can i use Rerun software ?
Hi @Mallak_Alqaisi,
For the Language-Directed Pick and Place demo, performance depends heavily on the system setup. Could you please confirm:
• Does stretch_system_check.py pass all checks?
• Are you running Ubuntu 22.04?
• What NVIDIA GPU are you using?
• Are you using a dedicated WiFi access point?
stretch_ai was primarily tested on Stretch 3 with a high-end NVIDIA GPU on Ubuntu 22.04, so mismatches can significantly affect results.
Best,
Jason
YES i passed all the checks test
I’m using Ubuntu 22.04
My PC specs: Graphics card: NIVIDIA(R) GeForce RTX™ 4090,24GB GDDR6X
Yes, I’m using a dedicated WiFi access point
Hi @Mallak_Alqaisi,
That’s a great starting point, your setup looks ideal, so we can focus on higher-level debugging now.
Before we proceed further, I’d like to double-check something important:
Stretch AI requires software to be installed both on the Stretch robot and on the separate GPU computer. The installation process is very important and must be completed carefully on both machines.
Could you please confirm that you followed the full installation guide here.
I just want to make sure the installation was completed properly before we continue troubleshooting, since even small deviations in the setup can affect performance significantly.
Also, when you can, please share:
- The exact command(s) you’re running
- What behavior you’re observing
- Any terminal output or error logs you’re seeing
Best,
Jason
1- Before I start trying docker, unplug the USB dongle on the robot
2- Run this command: sudo apt-get update
sudo apt-get install docker.io, then it requires stretch password
3- I use setup Docker Group So You Do Not Need To Use sudoand i got the screen below:
4- Clone the Stretch-AI Repository on your robot run this command: ./scripts/run_stretch_ai_ros2_bridge_server.sh and i got the screen below:
5- Run the Robot’s Script after i run this command to home the robot: stretch_robot_home.py
6- Run the Robot’s Script by run this command and then i got the screen below and is still running: ./scripts/run_stretch_ai_ros2_bridge_server.sh
7- Then i go to Create a virtual env on GPU Computer:
-
I use this command: mamba to see if mamba is installed on computer:
-
Activate mamba by rub this command: Run
mamba activate -
Then run those commands separately: Install Git LFS - needed for large files like images
sudo apt-get install git-lfs
git lfs installgit clone GitHub - hello-robot/stretch_ai --recursive
cd stretch_ai
./install.sh --cuda=$CUDA_VERSION --no-version
And i got these screen below:
** I arrived here, then i want to start doing Simple Installation Test, all of things are right now ?
Here i want to start Simple Installation Test:
1- let the GPU computer know the IP address Stretch robot by run this command: ./scripts/set_robot_ip.sh 192.168.1.2
So in this case do i need to Create a virtual env on your GPU Computer again?
After i try this Updating Stretch AI by run these command:
git pull -ff origin main
Optional; rarely needed
git submodule update --init --recursive
Then i got new environment: Environment name: stretch_ai_0.3.3
Python Version: 3.10 I don’t know what does it mean and why should i do that ??
Then I go to Simple Installation Test and run this command: ./scripts/set_robot_ip.sh 192.168.1.2
Then run this command: mamba activate stretch_ai_0.3.3
to get this (stretch_ai_0.3.3) hcmtlab@hcmtlab-Alienware-Aurora-ACT1250:~/stretch_ai$ THEN i run this command: python -m stretch.app.view_images
THEN I GIT THESE: the robot’s gripper open, the arm move, and then you see video from the robot’s cameras displayed on GPU computer AS BELOW
Hi @Mallak_Alqaisi,
From your latest results, it looks like you were able to successfully run the view_images application inside the Docker container on your GPU computer.
That’s excellent, this confirms that the GPU-side installation is working properly and communicating as expected.
If you have any specific questions or run into anything unexpected as you continue testing, please let us know and we’ll be happy to help.
Best,
Jason
When I’m trying to do language directed pick and place i did all steps but my question is do ineed to run these commands every single time:
Install Git LFS - needed for large files like images
sudo apt-get install git-lfs
git lfs install
Clone the repository
Do not forget the --recursive flag to clone submodules
git clone GitHub - hello-robot/stretch_ai --recursive
Run install script to create a conda environment and install dependencies
WARNING: this will delete an existing stretch_ai environment - do not do this to update your code!
cd stretch_ai
./install.sh --cuda=$CUDA_VERSION --no-version
because when i didn’t it, the demo didn’t work !
The screens below shown the results of run this command: python -m stretch.app.ai_pickup
with this error: AttributeError: ‘BaseModelOutputWithPooling’ object has no attribute ‘norm’
Is everything alright ? and if you have anything can help me with Rerun software
Hi @Mallak_Alqaisi,
You do not need to reinstall every time.
Those steps (Git LFS, clone, install.sh) are one-time setup only.
To run the language-directed pick & place pipeline, just do the following:
On the robot (from the stretch_ai folder):
stretch_robot_home.py
./scripts/run_stretch_ai_ros2_bridge_server.sh
On the GPU computer (from the stretch_ai folder):
mamba activate stretch_ai
./scripts/set_robot_ip.sh <robot_ip_address>
python -m stretch.app.view_images
That’s all that’s required to start the pipeline.
If it only worked after reinstalling, it’s likely the environment wasn’t activated or the robot bridge wasn’t running.
If it fails again, please share the exact error output and we’ll take a look.
Best,
Jason
(base) hcmtlab@hcmtlab-Alienware-Aurora-ACT1250:~$ mamba activate stretch_ai
./scripts/set_robot_ip.sh <192.168.1.2>
python -m stretch.app.view_images
bash: syntax error near unexpected token `newline’
/home/hcmtlab/mamba/envs/stretch_ai/bin/python: Error while finding module specification for ‘stretch.app.view_images’ (ModuleNotFoundError: No module named ‘stretch’)
This is what i got, in the same time i run the command on the Robot’s terminal and i note something as shown in the screen below, what does mean of version 0.3.3 ?
do in need to activate mamba stretch_ai_0.3.3 ?
Hi @Mallak_Alqaisi ,
You do not need to activate stretch_ai_0.3.3.
That 0.3.3 refers to the Docker image version on the robot side, not a mamba environment on the GPU computer.
From your robot logs, it looks like things are running correctly there.
Now on the GPU computer:
Activate mamba:
mamba activate stretch_ai
Set up the robot ip and run the app:
./scripts/set_robot_ip.sh 192.168.1.2
python -m stretch.app.view_images
Make sure you remove the < > around the IP address, they’re just placeholders.
Let me know if that works, and please share the GPU terminal output if it still errors.
Hi @Mallak_Alqaisi,
Could you confirm you are running the run_stretch_ai_ros2_bridge_server.sh script on the robot when you encounter this error (ModuleNotFoundError: No module named 'stretch')?
Also, could you share the full terminal output from running run_stretch_ai_ros2_bridge_server.sh?
That information will help us pinpoint the source of the issue.
Thank you.
Jason
Yes, I’m running the run_stretch_ai_ros2_bridge_server.sh script on the robot when i got this error (ModuleNotFoundError: No module named 'stretch')
Hi @Mallak_Alqaisi,
It looks like when you updated Stretch AI, the install script was run again without the --no-version flag, which created a new versioned environment: stretch_ai_0.3.3
To activate it, please run:
mamba activate stretch_ai_0.3.3
instead of:
mamba activate stretch_ai
Everything else looks correct.
Best,
Jason