RGB-D SLAM is the SLAM approach that combines color images with direct depth measurements to achieve dense and metric-consistent mapping - it simplifies geometric estimation compared with monocular methods by providing per-pixel range information.

What Is RGB-D SLAM?

• Definition: Localization and mapping pipeline using synchronized RGB and depth streams.
• Depth Source: Structured light, time-of-flight, or active stereo sensors.
• Output Types: Camera trajectory, dense surface map, and keyframe graph.
• Typical Environment: Indoor scenes with moderate range and texture.

Why RGB-D SLAM Matters

• Fast Geometry Access: Direct depth reduces triangulation uncertainty.
• Dense Mapping: Supports detailed surface reconstruction in real time.
• Robust Tracking: Combines appearance and geometry cues for pose estimation.
• AR and Robotics Utility: Strong for indoor navigation and interaction.
• Engineering Simplicity: Easier metric scale handling than monocular systems.

RGB-D SLAM Components

Pose Tracking:

• Align current RGB-D frame to map using geometric and photometric errors.
• Estimate incremental camera transform.

Map Fusion:

• Integrate depth observations into volumetric or surfel map.
• Maintain consistency across revisits.

Loop Closure:

• Detect revisited areas from visual descriptors.
• Correct drift with graph optimization.

How It Works

Step 1:

• Estimate frame-to-map pose using RGB features and depth alignment constraints.

Step 2:

• Fuse depth into global map and periodically run loop-closure optimization.

RGB-D SLAM is an efficient indoor mapping paradigm that pairs visual detail with direct depth for reliable metric reconstruction - it is a practical choice when depth sensors are available and operating conditions are suitable.