Glossary

RViz

RViz – definition

RViz is the standard 3D visualization tool used in ROS and ROS 2 to inspect robot state, sensor data, maps, coordinate frames, and planning outputs. In practical terms, RViz renders ROS message streams and transforms into an interactive scene. An engineer uses it to verify whether a mobile robot publishes correct geometry, whether localization is stable, whether a LiDAR scan is aligned with the base frame, and whether the navigation stack plans feasible motion.

In mobile robotics, RViz is not a simulator and not a perception algorithm. It is a visualization client. It subscribes to topics, reads the TF transform tree, and displays data such as sensor_msgs/LaserScan, sensor_msgs/PointCloud2, nav_msgs/OccupancyGrid, nav_msgs/Path, and visualization_msgs/Marker. In ROS 1, the package is commonly referred to as RViz. In ROS 2, the main application is RViz2, but the term RViz is still widely used in technical documentation.

For UGV platforms such as Leo Rover and Raph Rover, RViz is a primary debugging interface during bring-up, sensor integration, SLAM tests, and Nav2 configuration. It is especially important because these platforms are not autonomous out-of-the-box. A developer must validate every layer of the stack: odometry, TF, costmaps, localization, and command outputs.

How RViz works in ROS 2

RViz2 uses ROS 2 communication primitives to subscribe to topics and to render their content with a fixed reference frame. The fixed frame is usually map, odom, or base_link, depending on the debugging task. Correct TF data is essential. If transforms are missing, delayed, or inconsistent, RViz cannot place sensor data correctly in 3D space.

In a typical ROS 2 mobile robot setup, RViz depends on three information layers. First, robot geometry is loaded through URDF and robot_state_publisher. Second, dynamic transforms are published through TF2. Third, perception and navigation nodes publish maps, scans, point clouds, paths, and markers.

The ROS coordinate frame convention is defined in REP 103, while common mobile platform frame semantics are defined in REP 105. For a ground robot, the minimal useful frame chain is usually:

map -> odom -> base_link -> laser_frame
                         -> imu_link
                         -> camera_link

If that chain is broken, RViz may show classic symptoms such as floating scans, rotated point clouds, or a robot model that does not match measured motion.

Core displays used on UGV platforms

RViz supports many display types, but only a subset is critical for field and lab work on wheeled UGVs. The most useful displays are tied directly to ROS message types and TF diagnostics.

  • RobotModel – renders the URDF model described by /robot_description.
  • TF – shows the transform tree and frame axes.
  • LaserScan – visualizes 2D LiDAR data from sensor_msgs/LaserScan.
  • PointCloud2 – visualizes 3D LiDAR or depth camera output.
  • Map – displays occupancy grids from SLAM or localization stacks.
  • Path – shows global or local plans as nav_msgs/Path.
  • Odometry – displays pose and covariance from wheel odometry or fused state estimation.
  • Marker / MarkerArray – used by custom perception nodes for detections, clusters, regions of interest, and debugging geometry.

For Leo Rover, these displays are usually enough for validating a differential-drive ROS 2 stack on Raspberry Pi-based compute. For Raph Rover, the same workflow applies, but the larger payload budget often means more sensors, denser point clouds, and more demanding rendering.

Key parameters and metrics

RViz does not define robot performance, but it exposes parameters that indicate whether the system is configured correctly. Several measurable factors matter during integration.

Item Typical value or rule Why it matters
Fixed frame map, odom, or base_link Determines the global reference for all displays
LaserScan rate Often 5-20 Hz, sensor-dependent Low rate causes sparse updates, high rate increases CPU and DDS traffic
Point cloud rate Often 5-30 Hz, sensor-dependent Affects rendering load and network bandwidth
TF latency Should remain below sensor tolerance for stable overlay High latency causes scan-map misalignment
Occupancy grid resolution Commonly 0.02-0.10 m/cell Controls map detail and memory use
Image resolution Sensor-specific, often 640×480 or higher Impacts image display and transport load

These numbers come from the sensor and stack configuration, not from RViz itself. RViz is the place where errors become visible. For example, a 2D map at 0.05 m/cell may look stable, but if wheel odometry drifts or IMU orientation is wrong, the rendered scan overlay will reveal the issue quickly.

Example ROS 2 workflow

A common ROS 2 workflow is to start the robot description, state publisher, sensor drivers, SLAM or Nav2, and then open RViz2 with a predefined configuration. Saved RViz configuration files are important in repeatable lab work because they preserve displays, frame settings, and topic names.

ros2 run rviz2 rviz2
ros2 run rviz2 rviz2 -d config/nav2_debug.rviz

Before opening RViz, it is good practice to verify topic and transform availability.

ros2 topic list
ros2 topic hz /scan
ros2 topic echo /odom
ros2 run tf2_tools view_frames

A minimal launch pattern may look like this:

# conceptual structure
nodes:
  - robot_state_publisher
  - joint_state_publisher
  - lidar_driver
  - imu_driver
  - slam_toolbox
  - rviz2

Use cases with Leo Rover and Raph Rover

On Leo Rover, RViz is commonly used during first integration of a 2D LiDAR, depth camera, or GNSS receiver. Because the platform uses differential drive, engineers often inspect the relationship between /odom, wheel encoder output, and the base_link frame while driving on flat indoor or outdoor terrain. If Nav2 is added, RViz becomes the main interface for verifying global costmap, local costmap, inflation layer, and planned path.

On Raph Rover, the same tool supports heavier research payloads. Typical tasks include validating multi-sensor fusion, checking point cloud alignment from elevated sensor masts, and inspecting larger outdoor maps. A larger platform may carry more capable compute, but the integration logic remains the same: RViz confirms whether the TF tree, localization, and navigation outputs are geometrically consistent.

In both platforms, RViz is also useful for custom robotics services. When a team adds a camera-based detector, traversability estimator, or terrain segmentation node, the fastest way to validate output is often a MarkerArray overlay in RViz.

Limitations and trade-offs

RViz is a diagnostic interface, not a ground-truth validator. A display that looks correct does not guarantee metric accuracy. It only indicates consistency between published topics and transforms. Real validation still requires bag replay, benchmark datasets, hardware timing checks, and field tests.

There are also practical constraints. Rendering dense point clouds or multiple camera streams can overload embedded compute or remote desktop sessions. On Leo Rover, it is common to run heavy visualization on a remote workstation rather than on the onboard Raspberry Pi. On Raph Rover, local rendering is more realistic if the compute stack includes a stronger CPU or GPU, but bandwidth and DDS configuration still matter.

Normative references and standards

RViz usage in ROS-based mobile robotics should be understood together with the main ROS conventions and tools. The most relevant references are:

  • REP 103 – Standard Units of Measure and Coordinate Conventions.
  • REP 105 – Coordinate Frames for Mobile Platforms.
  • ROS 2 documentation – RViz2, TF2, URDF, and visualization message types.
  • nav_msgs, sensor_msgs, visualization_msgs – official ROS interface definitions.

When moving between ROS 1 and ROS 2, the conceptual role of RViz stays the same, but launch syntax, plugin packaging, and DDS-based communication in ROS 2 differ from ROS 1 transport and tooling.

See also

  • ROS 2
  • TF2
  • URDF
  • SLAM