> ## Documentation Index
> Fetch the complete documentation index at: https://innateinc-theo-docs-skills-authoring-api.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Foxglove Setup

> Visualize ROS2 topics from any machine using Foxglove.

Foxglove lets you see what your robot sees — camera feeds, LiDAR scans, navigation costmaps, TF frames, and `/cmd_vel` teleop — all from a browser. Works on macOS, Windows, and Linux.

<img src="https://mintcdn.com/innateinc-theo-docs-skills-authoring-api/9WIspYq4LkbGVthE/images/simulator/foxglove.png?fit=max&auto=format&n=9WIspYq4LkbGVthE&q=85&s=3ec84deba28d82e0adc09a8d640b0453" alt="Foxglove connected to MARS — camera feed, map, laser scan, and TF frames" width="2674" height="1742" data-path="images/simulator/foxglove.png" />

## How it works

MARS runs a **Foxglove Bridge** that exposes ROS2 topics over WebSocket. You connect to it from [Foxglove Studio](https://foxglove.dev/) on your laptop.

```
MARS (ROS2 topics) → Foxglove Bridge (WebSocket :8765) → Your browser
```

In the [simulator](/simulator/building-with-the-simulator#foxglove-and-ros-tooling) the bridge is always on. On a physical robot it is opt-in — you start it with the `innate` CLI.

***

## Setup

<Steps>
  <Step title="Start the bridge on MARS">
    SSH into the robot (or use a terminal on it) and start the bridge:

    ```bash theme={"languages":{"custom":["/languages/python-typed.json"]}}
    innate foxglove start    # prints ws://<robot-ip>:8765
    innate foxglove status   # is it running?
    innate foxglove stop     # stop it
    ```

    `innate foxglove start` prints the exact `ws://` URL to connect to.
  </Step>

  <Step title="Open Foxglove in your browser">
    Go to [app.foxglove.dev](https://app.foxglove.dev) (or open the desktop app).

    Click **Open connection** → **Foxglove WebSocket** → enter the printed URL:

    ```
    ws://<YOUR-ROBOT-IP>:8765
    ```
  </Step>

  <Step title="Add panels">
    Once connected, add visualization panels:

    * **Image** panel → select a camera topic (see below) for the camera feed
    * **3D** panel → see TF frames, LiDAR, point clouds, and navigation paths
    * **Plot** panel → graph any numeric topic over time
    * **Raw Messages** panel → inspect any topic's raw data
  </Step>
</Steps>

<Note>
  **Over Wi-Fi, subscribe to the `/mars/main_camera/remote/*` topics, not the raw ones.** The raw camera images (\~0.9 MB/frame) and point clouds (`/mars/main_camera/points`, \~10 MB/s) are far more than a Wi-Fi link can carry, so every panel — including `/cmd_vel` — falls seconds behind. The `remote/` namespace carries the same topics throttled to \~2 Hz (images already compressed), which fits comfortably. Prefer the `.../compressed` image topics and mono `remote/points` over `remote/points_color`. On a wired connection or the simulator, the raw topics are fine.
</Note>

***

## Useful topics to visualize

| Topic                                                       | Type            | What you see                                       |
| ----------------------------------------------------------- | --------------- | -------------------------------------------------- |
| `/mars/main_camera/left/image_rect_color`                   | Image           | Main camera feed (rectified color)                 |
| `/mars/main_camera/left/image_rect_color/compressed`        | CompressedImage | Same feed, compressed — prefer over Wi-Fi          |
| `/mars/main_camera/remote/left/image_rect_color/compressed` | CompressedImage | Throttled camera for Wi-Fi                         |
| `/mars/main_camera/points`                                  | PointCloud2     | Depth point cloud (use `remote/points` over Wi-Fi) |
| `/scan`                                                     | LaserScan       | LiDAR sweep                                        |
| `/odom`                                                     | Odometry        | Robot position and velocity                        |
| `/cmd_vel`                                                  | Twist           | Velocity commands being sent                       |
| `/map`                                                      | OccupancyGrid   | SLAM / navigation map                              |
| `/battery_state`                                            | BatteryState    | Battery voltage and percentage                     |
| `/tf`, `/tf_static`                                         | TF              | All coordinate frames                              |

***

## RViz alternative (Linux only)

If you're on a Linux machine with ROS2 Humble installed, you can use RViz directly. It connects to MARS over Zenoh DDS — no bridge needed.

Make sure your machine is on the same network and Zenoh discovery can reach the robot. Then:

```bash theme={"languages":{"custom":["/languages/python-typed.json"]}}
source /opt/ros/humble/setup.bash
rviz2
```

Add displays for the topics you care about. RViz gives you the most complete visualization experience but requires a native Linux + ROS2 setup.

***

## Troubleshooting

<AccordionGroup>
  <Accordion title="Can't connect to WebSocket">
    * Verify MARS and your laptop are on the same network.
    * Check the bridge is running: `innate foxglove status`, or `ros2 node list | grep foxglove`.
    * Use the exact `ws://` URL that `innate foxglove start` printed.
    * Try the IP address directly instead of a `.local` hostname.
    * Make sure port `8765` isn't blocked by a firewall.
  </Accordion>

  <Accordion title="No topics showing up">
    * The bridge only exposes topics that are actively published. Make sure ROS nodes are running: `innate view`.
    * Try `ros2 topic list` on the robot to confirm topics exist.
  </Accordion>

  <Accordion title="High latency on camera feed">
    * Image topics are bandwidth-heavy. Over Wi-Fi, use the `/mars/main_camera/remote/*` topics and `.../compressed` variants (see the note above); for the full-rate raw topics, use a wired connection.
    * In Foxglove, reduce the image panel's update rate.
  </Accordion>
</AccordionGroup>