Skip to content

xinxun-robot/dora-visualization

BranchesTags

Repository files navigation

DViz

A modern robotics visualization tool built in Rust, combining Makepad UI, Rerun 3D visualization, and Zenoh distributed networking for debugging robots and autonomous systems.

DViz Demo

DViz Screenshot

Features

  • Distributed-first architecture: Debug robots remotely over LAN using Zenoh auto-discovery
  • Dora dataflow integration: Visualize sensor pipelines and node activity
  • Rerun 3D visualization: Point clouds, transforms, markers, robot models
  • ROS bag playback: Multi-sensor support (LiDAR, IMU, GPS, TF)
  • URDF robot models: Load and visualize robot meshes

Installation

Prerequisites

# Rust (nightly required for Makepad)
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
rustup install nightly
rustup default nightly

# Dora CLI
cargo install dora-cli

# Python dependencies (for example dataflows)
pip install dora-rs pyyaml numpy pyarrow

Build DViz

git clone https://github.com/bobd988/dviz.git
cd dviz
cargo build --release

This builds:

  • dviz - Main visualization shell
  • dviz-dora-bridge - Dora node that publishes data via Zenoh

Quick Start (Single Machine)

# Terminal 1: Start Dora daemon
dora up

# Terminal 2: Run a dataflow
dora start dataflow.yml --name dviz

# Terminal 3: Run DViz
cargo run -p dviz-shell --release

In the DViz UI:

  1. Click Spawn Rerun to open the 3D viewer
  2. Zenoh auto-connects on startup (status shows "Zenoh: Connected")
  3. View vehicle simulation data in Rerun

Example: Path Following Simulation

This example runs a vehicle path-following simulation with:

  • bicycle_model - Vehicle physics simulator
  • simple_planner - Pure pursuit path following
  • imu_synthesizer - Synthetic IMU generation
  • dviz_bridge - Publishes data via Zenoh

Run Locally

# Terminal 1
dora up

# Terminal 2
dora start examples/dataflow-path-following.yml --name pathfollow

# Terminal 3
cargo run -p dviz-shell --release

View in Rerun

The visualization shows:

  • Vehicle pose: Arrow showing position and heading
  • Waypoints: Line strip showing planned path
  • Target point: Current pursuit target
  • IMU data: Acceleration/gyroscope scalars

Example: Point Cloud Mapping

This example runs a vehicle mapping pipeline that builds a 3D map from point cloud data using ICP odometry.

Prerequisites

# Additional Python dependencies
pip install open3d

# Clone dora-examples (contains the mapping operators)
git clone https://github.com/dora-rs/dora-examples.git ~/dora-examples

Run Mapping Dataflow

# Terminal 1: Start Dora daemon
dora up

# Terminal 2: Run the mapping dataflow
dora start examples/dataflow-mapping.yml --name mapping

# Terminal 3: Run DViz
cargo run -p dviz-shell --release

View in Rerun

In the DViz UI:

  1. Click Spawn Rerun to open the 3D viewer
  2. Zenoh auto-connects and receives mapping data

The visualization shows:

  • Point clouds: LiDAR scans from PCD files
  • Odometry pose: Vehicle position estimated by ICP
  • Trajectory: Path traveled by the vehicle

Example: ROS Bag Playback

DViz can play back ROS1 bag files with multi-sensor visualization support.

Supported Sensors

  • PointCloud2: LiDAR point clouds (e.g., Velodyne)
  • Imu: Accelerometer and gyroscope data
  • NMEA Sentence: GPS position from NMEA messages
  • TimeReference: GPS time synchronization
  • Temperature: Temperature sensor readings

Play a ROS Bag

# Terminal 1: Run DViz
cargo run -p dviz-shell --release

In the DViz UI:

  1. Click Spawn Rerun to open the 3D viewer
  2. Click File → Open Bag (or use the File button)
  3. Select the bag file: examples/hdl_400.bag
  4. Click Play to start playback

Playback Controls

  • Play/Pause: Toggle playback
  • Stop: Reset to beginning
  • Seek: Jump to specific time in the bag

Visualization in Rerun

The hdl_400.bag example displays:

  • world/lidar: Velodyne point cloud (height-colored)
  • world/imu/accel_arrow: Acceleration vector (cyan)
  • world/imu/gyro_arrow: Angular velocity vector (orange)
  • world/imu/accel_x,y,z: Acceleration scalars
  • world/imu/gyro_x,y,z: Gyroscope scalars
  • world/gps/position: GPS position (green point)
  • world/gps/latitude,longitude,altitude: GPS scalars
  • world/temperature/celsius: Temperature reading

Distributed Setup (Robot + Debug PC)

DViz is designed for remote debugging where:

  • Robot runs the dataflow headlessly (no display)
  • Debug PC runs DViz to visualize robot data over LAN

Architecture

┌─────────────────────┐         Zenoh (LAN)         ┌─────────────────────┐
│       ROBOT         │ ─────────────────────────── │      DEBUG PC       │
│  (192.168.1.10)     │         UDP Multicast       │   (192.168.1.20)    │
│                     │         224.0.0.225:7446    │                     │
│  dora up            │                             │  cargo run -p       │
│  dora start         │                             │    dviz-shell       │
│    dataflow-robot   │         ────────────>       │                     │
│      .yml           │         sensor data         │  Rerun 3D viewer    │
└─────────────────────┘                             └─────────────────────┘

Step 1: Robot Side (Headless)

SSH into the robot and run:

cd /path/to/dviz

# Build the bridge (first time only)
cargo build --release -p dviz-rerun-bridge

# Start Dora daemon
dora up

# Run the robot dataflow
dora start examples/dataflow-robot.yml --name robot

The robot will publish data to Zenoh topics:

  • dviz/pose - Vehicle position
  • dviz/state - Vehicle state
  • dviz/imu - IMU readings
  • dviz/waypoints - Planned path

Step 2: Debug PC Side

cd /path/to/dviz

# Run DViz
cargo run -p dviz-shell --release

DViz auto-connects via Zenoh multicast scouting. Both machines must be on the same LAN subnet.

Zenoh Network Configuration

Default: Auto-Discovery (Recommended)

Zenoh uses multicast scouting by default:

  • UDP multicast on 224.0.0.225:7446
  • Works automatically on the same LAN subnet
  • No configuration needed

When Auto-Discovery Fails

If machines are on different subnets or multicast is blocked:

Option 1: Direct Connection

Set environment variable on the subscriber (debug PC):

ZENOH_CONNECT=tcp/192.168.1.10:7447 cargo run -p dviz-shell

Or in the dataflow YAML on the robot:

nodes:
  - id: dviz_bridge
    path: target/release/dviz-dora-bridge
    env:
      ZENOH_CONNECT: tcp/192.168.1.20:7447  # Debug PC address

Option 2: Zenoh Router

Run a Zenoh router as a central hub:

# On any machine accessible by both robot and PC
zenohd

Then configure both sides to connect:

# Robot
ZENOH_CONNECT=tcp/router-ip:7447 dora start dataflow-robot.yml

# PC
ZENOH_CONNECT=tcp/router-ip:7447 cargo run -p dviz-shell

When IP Addresses Change

Scenario 1: Robot IP Changed

  1. Find new robot IP: ip addr or hostname -I
  2. If using direct connection, update debug PC: 
    ZENOH_CONNECT=tcp/<new-robot-ip>:7447 cargo run -p dviz-shell
  3. If using auto-discovery, no changes needed (same subnet)

Scenario 2: Debug PC IP Changed

  1. If robot is configured to connect to PC: 
    # Update dataflow-robot.yml
env:
  ZENOH_CONNECT: tcp/<new-pc-ip>:7447
  2. Restart the dataflow: 
    dora stop robot
dora start examples/dataflow-robot.yml --name robot

Scenario 3: Different Subnet/Network

Use a Zenoh router with a static IP that both can reach:

# Router machine (e.g., 192.168.1.1)
zenohd --listen tcp/0.0.0.0:7447

# Robot (any subnet)
ZENOH_CONNECT=tcp/192.168.1.1:7447 dora start ...

# PC (any subnet)
ZENOH_CONNECT=tcp/192.168.1.1:7447 cargo run -p dviz-shell

Network Troubleshooting

Symptom Cause Solution
"Zenoh: Connecting..." hangs Multicast blocked Use ZENOH_CONNECT
No data in Rerun Firewall blocking Open UDP 7446, TCP 7447
Intermittent data Network congestion Use wired connection
Works locally, not remote Different subnets Use Zenoh router

Firewall Configuration

# Linux (robot and PC)
sudo ufw allow 7446/udp  # Zenoh scouting
sudo ufw allow 7447/tcp  # Zenoh data

# Or disable firewall for testing
sudo ufw disable

Environment Variables

Variable Default Description
ZENOH_CONNECT (auto-discovery) Router address, e.g., tcp/192.168.1.100:7447
ZENOH_TOPIC_PREFIX dviz Topic namespace for all messages
RUST_LOG info Log level (debug, info, warn, error)

Dataflow Examples

File Description
dataflow.yml Local simulation with DViz bridge
examples/dataflow-path-following.yml Vehicle path following (same as above)
examples/dataflow-robot.yml Headless robot dataflow for remote debug
examples/dataflow-ros2.yml ROS2 bridge integration
examples/dataflow-mapping.yml Point cloud mapping pipeline

Project Structure

dviz/
├── dviz-core/           # Core types (Transform, PointCloud, Marker)
├── dviz-transform/      # Coordinate frame system
├── dviz-displays/       # Display plugins (Grid, TF, PointCloud, etc.)
├── dviz-urdf/           # URDF parsing and robot models
├── dviz-shell/          # Main application (Makepad UI)
├── dviz-widgets/        # Reusable UI components
├── dviz-rerun-bridge/   # Dora-to-Zenoh bridge node
├── dviz-rosbag/         # ROS bag playback
├── examples/            # Dataflow configurations
└── docs/                # Design documentation

Zenoh Message Protocol

DViz uses a universal JSON protocol that any application can publish:

{
  "type": "points3d",
  "timestamp": 1.5,
  "data": {
    "positions": [[0, 0, 0], [1, 1, 1]],
    "color": [255, 0, 0, 255]
  }
}

Supported types: points3d, arrows3d, linestrips3d, boxes3d, transform3d, scalar, image, log

Topic format: {prefix}/{entity_path} (e.g., dviz/world/vehicle/lidar)

License

Apache-2.0

About

rerun based dora robot tools

Resources

Readme
Activity
Custom properties

Stars

1 star

Watchers

0 watching

Forks

0 forks
Report repository

Releases

40 tags

Packages

 
 
 

Contributors

Languages