chore: address PR feedback

2025-04-09 13:44:07 +02:00 · 2025-04-09 13:44:07 +02:00 · c586de3e02
parent c197bd44a9
commit c586de3e02
12 changed files with 33 additions and 50 deletions
--- a/README.md
+++ b/README.md
@ -103,6 +103,11 @@ When using `miniconda`, install `ffmpeg` in your environment:
 conda install ffmpeg -c conda-forge
 ```

+> **NOTE:** This usually installs `ffmpeg 7.X` for your platform (check the version installed with `ffmpeg -encoders | grep libsvtav1`). If it isn't `ffmpeg 7.X` or lacks `libsvtav1` support, you can explicitly install `ffmpeg 7.X` using:
+```
+conda install ffmpeg=7.1.1 -c conda-forge
+```
+
 Install 🤗 LeRobot:
 ```bash
 pip install -e .
--- a/benchmarks/video/capture_camera_feed.py
+++ b/benchmarks/video/capture_camera_feed.py
@ -24,11 +24,13 @@ from pathlib import Path
 import cv2
 import rerun as rr

+# see https://rerun.io/docs/howto/visualization/limit-ram
+RERUN_MEMORY_LIMIT = os.getenv("LEROBOT_RERUN_MEMORY_LIMIT", "5%")
+

 def display_and_save_video_stream(output_dir: Path, fps: int, width: int, height: int, duration: int):
    rr.init("lerobot_capture_camera_feed")
-    memory_limit = os.getenv("LEROBOT_RERUN_MEMORY_LIMIT", "5%")
-    rr.spawn(memory_limit=memory_limit)
+    rr.spawn(memory_limit=RERUN_MEMORY_LIMIT)

    now = dt.datetime.now()
    capture_dir = output_dir / f"{now:%Y-%m-%d}" / f"{now:%H-%M-%S}"
@ -57,10 +59,10 @@ def display_and_save_video_stream(output_dir: Path, fps: int, width: int, height
        cv2.imwrite(str(capture_dir / f"frame_{frame_index:06d}.png"), frame)
        frame_index += 1

-    # Release the capture and destroy all windows
+    # Release the capture
    cap.release()
-    # TODO(Steven): Find a way to close visualizer: https://github.com/rerun-io/rerun/pull/9400
-    # cv2.destroyAllWindows()
+
+    # TODO(Steven): Add a graceful shutdown via a close() method for the Viewer context, though not currently supported in the Rerun API.


 if __name__ == "__main__":
--- a/examples/10_use_so100.md
+++ b/examples/10_use_so100.md
@ -498,11 +498,7 @@ python lerobot/scripts/control_robot.py \
 #### a. Teleop with displaying cameras
 Follow [this guide to setup your cameras](https://github.com/huggingface/lerobot/blob/main/examples/7_get_started_with_real_robot.md#c-add-your-cameras-with-opencvcamera). Then you will be able to display the cameras on your computer while you are teleoperating by running the following code. This is useful to prepare your setup before recording your first dataset.

-> **NOTE:** To visualize the data, enable `--control.display_data=true`. This streams the data using `rerun`. You can adjust the viewer's behavior with these environment variables (refer to `rerun` documentation for more information):
-> - [`RERUN_FLUSH_NUM_BYTES`](https://rerun.io/docs/reference/sdk/micro-batching) [default: 8000]
-> - [`LEROBOT_RERUN_MEMORY_LIMIT`](https://rerun.io/docs/howto/visualization/limit-ram) [default: 5%]
-> - [`LEROBOT_VIEWER_IP`](https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp) (only for remote robots) [default: None]
-> - [`LEROBOT_VIEWER_PORT`](https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp) (only for remote robots) [default: 9876]
+> **NOTE:** To visualize the data, enable `--control.display_data=true`. This streams the data using `rerun`.

 ```bash
 python lerobot/scripts/control_robot.py \
--- a/examples/11_use_lekiwi.md
+++ b/examples/11_use_lekiwi.md
@ -424,11 +424,7 @@ python lerobot/scripts/control_robot.py \
  --control.fps=30
 ```

-> **NOTE:** To visualize the data, enable `--control.display_data=true`. This streams the data using `rerun`. You can adjust the viewer's behavior with these environment variables (refer to `rerun` documentation for more information):
-> - [`RERUN_FLUSH_NUM_BYTES`](https://rerun.io/docs/reference/sdk/micro-batching) [default: 8000]
-> - [`LEROBOT_RERUN_MEMORY_LIMIT`](https://rerun.io/docs/howto/visualization/limit-ram) [default: 5%]
-> - [`LEROBOT_VIEWER_IP`](https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp) (only for remote robots) [default: None]
-> - [`LEROBOT_VIEWER_PORT`](https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp) (only for remote robots) [default: 9876]
+> **NOTE:** To visualize the data, enable `--control.display_data=true`. This streams the data using `rerun`. For the `--control.type=remote_robot` you will also need to set `--control.viewer_ip` and `--control.viewer_port`

 You should see on your laptop something like this: ```[INFO] Connected to remote robot at tcp://172.17.133.91:5555 and video stream at tcp://172.17.133.91:5556.``` Now you can move the leader arm and use the keyboard (w,a,s,d) to drive forward, left, backwards, right. And use (z,x) to turn left or turn right. You can use (r,f) to increase and decrease the speed of the mobile robot. There are three speed modes, see the table below:
 | Speed Mode | Linear Speed (m/s) | Rotation Speed (deg/s) |
--- a/examples/11_use_moss.md
+++ b/examples/11_use_moss.md
@ -219,11 +219,7 @@ python lerobot/scripts/control_robot.py \
 **Teleop with displaying cameras**
 Follow [this guide to setup your cameras](https://github.com/huggingface/lerobot/blob/main/examples/7_get_started_with_real_robot.md#c-add-your-cameras-with-opencvcamera). Then you will be able to display the cameras on your computer while you are teleoperating by running the following code. This is useful to prepare your setup before recording your first dataset.

-> **NOTE:** To visualize the data, enable `--control.display_data=true`. This streams the data using `rerun`. You can adjust the viewer's behavior with these environment variables (refer to `rerun` documentation for more information):
-> - [`RERUN_FLUSH_NUM_BYTES`](https://rerun.io/docs/reference/sdk/micro-batching) [default: 8000]
-> - [`LEROBOT_RERUN_MEMORY_LIMIT`](https://rerun.io/docs/howto/visualization/limit-ram) [default: 5%]
-> - [`LEROBOT_VIEWER_IP`](https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp) (only for remote robots) [default: None]
-> - [`LEROBOT_VIEWER_PORT`](https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp) (only for remote robots) [default: 9876]
+> **NOTE:** To visualize the data, enable `--control.display_data=true`. This streams the data using `rerun`.

 ```bash
 python lerobot/scripts/control_robot.py \
--- a/examples/7_get_started_with_real_robot.md
+++ b/examples/7_get_started_with_real_robot.md
@ -56,11 +56,7 @@ Now you are ready to configure your motors for the first time, as detailed in th

 If you have already configured your motors the first time, you can streamline the process by directly running the teleoperate script (which is detailed further in the tutorial):

-> **NOTE:** To visualize the data, enable `--control.display_data=true`. This streams the data using `rerun`. You can adjust the viewer's behavior with these environment variables (refer to `rerun` documentation for more information):
-> - [`RERUN_FLUSH_NUM_BYTES`](https://rerun.io/docs/reference/sdk/micro-batching) [default: 8000]
-> - [`LEROBOT_RERUN_MEMORY_LIMIT`](https://rerun.io/docs/howto/visualization/limit-ram) [default: 5%]
-> - [`LEROBOT_VIEWER_IP`](https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp) (only for remote robots) [default: None]
-> - [`LEROBOT_VIEWER_PORT`](https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp) (only for remote robots) [default: 9876]
+> **NOTE:** To visualize the data, enable `--control.display_data=true`. This streams the data using `rerun`.

 ```bash
 python lerobot/scripts/control_robot.py \
@ -834,6 +830,10 @@ It contains:
 - `dtRphone:33.84 (29.5hz)` which is the delta time of capturing an image from the phone camera in the thread running asynchronously.

 Troubleshooting:
+- On Linux, if you encounter any issue during video encoding with `ffmpeg: unknown encoder libsvtav1`, you can:
+   - install with conda-forge by running `conda install -c conda-forge ffmpeg` (it should be compiled with `libsvtav1`),
+   - or, install [ffmpeg build dependencies](https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu#GettheDependencies) and [compile ffmpeg from source with libsvtav1](https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu#libsvtav1),
+   - and, make sure you use the corresponding ffmpeg binary to your install with `which ffmpeg`.
 - On Linux, if the left and right arrow keys and escape key don't have any effect during data recording, make sure you've set the `$DISPLAY` environment variable. See [pynput limitations](https://pynput.readthedocs.io/en/latest/limitations.html#linux).

 At the end of data recording, your dataset will be uploaded on your Hugging Face page (e.g. https://huggingface.co/datasets/cadene/koch_test) that you can obtain by running:
--- a/examples/8_use_stretch.md
+++ b/examples/8_use_stretch.md
@ -103,12 +103,7 @@ Before trying teleoperation, you need activate the gamepad controller by pressin

 Now try out teleoperation (see above documentation to learn about the gamepad controls):

-> **NOTE:** To visualize the data, enable `--control.display_data=true`. This streams the data using `rerun`. You can adjust the viewer's behavior with these environment variables (refer to `rerun` documentation for more information):
-> - [`RERUN_FLUSH_NUM_BYTES`](https://rerun.io/docs/reference/sdk/micro-batching) [default: 8000]
-> - [`LEROBOT_RERUN_MEMORY_LIMIT`](https://rerun.io/docs/howto/visualization/limit-ram) [default: 5%]
-> - [`LEROBOT_VIEWER_IP`](https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp) (only for remote robots) [default: None]
-> - [`LEROBOT_VIEWER_PORT`](https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp) (only for remote robots) [default: 9876]
-
+> **NOTE:** To visualize the data, enable `--control.display_data=true`. This streams the data using `rerun`.
 ```bash
 python lerobot/scripts/control_robot.py \
    --robot.type=stretch \
--- a/examples/9_use_aloha.md
+++ b/examples/9_use_aloha.md
@ -49,11 +49,7 @@ Teleoperation consists in manually operating the leader arms to move the followe

 By running the following code, you can start your first **SAFE** teleoperation:

-> **NOTE:** To visualize the data, enable `--control.display_data=true`. This streams the data using `rerun`. You can adjust the viewer's behavior with these environment variables (refer to `rerun` documentation for more information):
-> - [`RERUN_FLUSH_NUM_BYTES`](https://rerun.io/docs/reference/sdk/micro-batching) [default: 8000]
-> - [`LEROBOT_RERUN_MEMORY_LIMIT`](https://rerun.io/docs/howto/visualization/limit-ram) [default: 5%]
-> - [`LEROBOT_VIEWER_IP`](https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp) (only for remote robots) [default: None]
-> - [`LEROBOT_VIEWER_PORT`](https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp) (only for remote robots) [default: 9876]
+> **NOTE:** To visualize the data, enable `--control.display_data=true`. This streams the data using `rerun`.

 ```bash
 python lerobot/scripts/control_robot.py \
--- a/lerobot/common/robot_devices/control_configs.py
+++ b/lerobot/common/robot_devices/control_configs.py
@ -118,6 +118,9 @@ class RemoteRobotConfig(ControlConfig):
    log_interval: int = 100
    # Display all cameras on screen
    display_data: bool = False
+    # Rerun configuration for remote robot (https://ref.rerun.io/docs/python/0.22.1/common/initialization_functions/#rerun.connect_tcp)
+    viewer_ip: str | None = None
+    viewer_port: str | None = None


@dataclass
--- a/lerobot/common/robot_devices/control_utils.py
+++ b/lerobot/common/robot_devices/control_utils.py
@ -307,10 +307,6 @@ def stop_recording(robot, listener, display_data):
    if not is_headless() and listener is not None:
        listener.stop()

-        # TODO(Steven): Find a way to close visualizer: https://github.com/rerun-io/rerun/pull/9400
-        # if display_data:
-        #     cv2.destroyAllWindows()
-

 def sanity_check_dataset_name(repo_id, policy_cfg):
    _, dataset_name = repo_id.split("/")
--- a/lerobot/scripts/control_robot.py
+++ b/lerobot/scripts/control_robot.py
@ -381,26 +381,24 @@ def _init_rerun(control_config: ControlConfig, session_name: str = "lerobot_cont
    if (control_config.display_data and not is_headless()) or (
        control_config.display_data and isinstance(control_config, RemoteRobotConfig)
    ):
-        # Configure Rerun flush batch size
+        # Configure Rerun flush batch size default to 8KB if not set
        batch_size = os.getenv("RERUN_FLUSH_NUM_BYTES", "8000")
        os.environ["RERUN_FLUSH_NUM_BYTES"] = batch_size

-        # Get memory limit and viewer connection parameters
-        memory_limit = os.getenv("LEROBOT_RERUN_MEMORY_LIMIT", "5%")
-        viewer_ip = os.getenv("LEROBOT_VIEWER_IP")
-        viewer_port = os.getenv("LEROBOT_VIEWER_PORT", "9876")
-
        # Initialize Rerun based on configuration
        rr.init(session_name)
        if isinstance(control_config, RemoteRobotConfig):
-            if not viewer_ip:
+            viewer_ip = control_config.viewer_ip
+            viewer_port = control_config.viewer_port
+            if not viewer_ip or not viewer_port:
                raise ValueError(
-                    "Viewer IP required for remote config. Set LEROBOT_VIEWER_IP "
-                    "or disable control_config.display_data."
+                    "Viewer IP & Port are required for remote config. Set via config file/CLI or disable control_config.display_data."
                )
            logging.info(f"Connecting to viewer at {viewer_ip}:{viewer_port}")
            rr.connect_tcp(f"{viewer_ip}:{viewer_port}")
        else:
+            # Get memory limit for rerun viewer parameters
+            memory_limit = os.getenv("LEROBOT_RERUN_MEMORY_LIMIT", "10%")
            rr.spawn(memory_limit=memory_limit)


--- a/pyproject.toml
+++ b/pyproject.toml
@ -62,7 +62,7 @@ dependencies = [
    "omegaconf>=2.3.0",
    "opencv-python-headless>=4.9.0",
    "packaging>=24.2",
-    "av>=14.0.0",
+    "av>=12.0.5",
    "pymunk>=6.6.0",
    "pynput>=1.7.7",
    "pyzmq>=26.2.1",