TimeoutError: Timed out waiting for frame from camera OpenCVCamera(0) after 200 ms. Read thread alive: True

[CrossyNautilus]

System Info

LeRobot version: f771e3eaf1b30aae65768513b2e67d351f5f6c2f
OS: MacOS Ventura 13.3.1
Python version: 3.10.18
Numpy version: 2.2.6
Torch version: 2.7.1
LeRobot configuration:
`lerobot` version: 0.3.2
- Platform: macOS-13.3.1-arm64-arm-64bit
- Python version: 3.10.18
- Huggingface_hub version: 0.34.3
- Dataset version: 3.6.0
- Numpy version: 2.2.6
- PyTorch version (GPU?): 2.7.1 (False)
- Cuda version: N/A
- Using GPU in script?: <fill in>

Information

• One of the scripts in the examples/ folder of LeRobot
• My own task or dataset (give details below)

Reproduction

python -m lerobot.record
--robot.type=so100_follower
--robot.port=/dev/tty.usbmodem58CD1770331
--robot.id=my_awesome_follower_arm
--robot.cameras="{ wrist.right: {type: opencv, index_or_path: 0, width: 1920, height: 1080, fps: 30}, top: {type: opencv, index_or_path: 1, width: 1920, height: 1080, fps: 30}}"
--teleop.type=so100_leader
--teleop.port=/dev/tty.usbmodem58CD1770511
--teleop.id=my_awesome_leader_arm
--display_data=true
--dataset.repo_id=${HF_USER}/record-test
--dataset.num_episodes=5
--dataset.single_task="Stack the yellow birdie"

Expected behavior

I expect to be able to record a dataset, but it randomly disconnects between 5-20 seconds after running the script above and displays the error message.

[linshenghou]

---

Step 1: Check Your Camera Formats

v4l2-ctl --device=/dev/video0 --list-formats-ext
v4l2-ctl --device=/dev/video1 --list-formats-ext

Look for the output - this is what we need.

Step 2: Use MJPEG Format in Your Command

check this pr #1558

python -m lerobot.record \
--robot.type=so100_follower \
--robot.port=/dev/tty.usbmodem58CD1770331 \
--robot.id=my_awesome_follower_arm \
--robot.cameras="{ wrist.right: {type: opencv, index_or_path: 0, width: 1920, height: 1080, fps: 30, fourcc: 'MJPG'}, top: {type: opencv, index_or_path: 1, width: 1920, height: 1080, fps: 30, fourcc: 'MJPG'}}" \
--teleop.type=so100_leader \
--teleop.port=/dev/tty.usbmodem58CD1770511 \
--teleop.id=my_awesome_leader_arm \
--display_data=true \
--dataset.repo_id=${HF_USER}/record-test \
--dataset.num_episodes=5 \
--dataset.single_task="Stack the yellow birdie"

The key change: Added fourcc: 'MJPG' to both cameras.

Why This Fixes It:

• Your cameras support MJPEG (check with v4l2-ctl)
• MJPEG is much more stable for high resolution

That's it! Try this and let me know if it works.

[CrossyNautilus]

Thanks for the reply!

v4l2-ctl --device=/dev/video0 --list-formats-ext

I don't have v4l2-ctl installed on my Mac, how can I work around this?

When I run your lerobot.record command, I get the following error:

draccus.utils.DecodingError: robot.cameras.wrist.right: The fields fourcc are not valid for OpenCVCameraConfig

Let me know what else I can try!

[linshenghou]

---

Step 1: u can try this

lerobot-find-cameras opencv

Look for the output, check ur camera supporting format.

Step 2: pls check pr first, that's a demo command

check this pr #1558

The key change: Added fourcc: 'MJPG' to both cameras.

That's it! Try this and let me know if it works.

[CrossyNautilus]

Thanks for your help, I was able to get the episode to record without timeout errors. When do you plan to merge the PR?

[linshenghou]

good for u!

[Liu-Xiaoyan97]

Thanks for your help, I was able to get the episode to record without timeout errors. When do you plan to merge the PR?

How do you solve this problem? I'm using a mac mini M4. There are a total of 7 usb ports, three of which are for the camera. Their connection methods are similar to
hub-1
| - camera 1
| - camera 2
| - hub2
| -- | - right leader
| -- | - left leader
| -- |- hub 3
| -- -- | - right follower
| -- -- | - left follower
| -- -- | - camera 3

I can read at most two cameras simultaneously. The initialization of the third camera was successful, but the reading failed.

[Dionaea]

Thanks for your help, I was able to get the episode to record without timeout errors. When do you plan to merge the PR?

How do you solve this problem? I'm using a mac mini M4. There are a total of 7 usb ports, three of which are for the camera. Their connection methods are similar to hub-1 | - camera 1 | - camera 2 | - hub2 | -- | - right leader | -- | - left leader | -- |- hub 3 | -- -- | - right follower | -- -- | - left follower | -- -- | - camera 3

I can read at most two cameras simultaneously. The initialization of the third camera was successful, but the reading failed.

Got the same problem on macbook air m3. The solution was to move all the cameras to hub1

[jzchen2]

I am getting this same issue, what was the solution after to the error saying "fourcc: field not valid"?

[at1047]

@jzchen2 u can hardcode the fourcc option in the code, since the python file doesn't take it as input, but you need it for not to timeout

[jzchen2]

@at1047 Would you happen to know which python file it is exactly? I tried looking for the file and where to add the argument but the robot never worked after I added it, so I assume I added it in the wrong spot.

[linshenghou]

after the Lerobot released its new version with the pr #1558 been merged, i think u can directly try command like

python -m lerobot.record \
--robot.type=so100_follower \
--robot.port=/dev/tty.XXXX \
--robot.id=my_awesome_follower_arm \
--robot.cameras="{ wrist.right: {type: opencv, index_or_path: 0, width: 1920, height: 1080, fps: 30, fourcc: 'MJPG'}, top: {type: opencv, index_or_path: 1, width: 1920, height: 1080, fps: 30, fourcc: 'MJPG'}}" \
--teleop.type=so100_leader \
--teleop.port=/dev/tty.XXXX \
--teleop.id=my_awesome_leader_arm \
--display_data=true \
--dataset.repo_id=${HF_USER}/record-test \
--dataset.num_episodes=5 \
--dataset.single_task="Stack the yellow birdie"