← Wiki

Ingredient

RGB-D camera

Also known as: depth camera, Intel RealSense, Stereolabs ZED, Orbbec Astra

Camera that captures color (RGB) and per-pixel depth (D) simultaneously — produces a 3D point cloud at video frame rates. The right ingredient for any robot that needs to perceive 3D structure: pick-and-place arms, autonomous rovers navigating uneven ground, fruit-detection systems that need both visual ID and reach distance. Three principal technologies: structured-light (older Kinect, Astra), active stereo (Intel RealSense D435/D455), passive stereo (ZED). USB 3.0 typical. ~$200–500 hobbyist; $800+ industrial.

Inputs / outputs

  • Color stream: typically 1080p @ 30 FPS or 720p @ 60 FPS
  • Depth stream: 480p–720p @ 30–90 FPS, range 0.3–10 m typical
  • IMU: many models include 6-axis IMU (RealSense D435i)
  • Interface: USB 3.0 (USB 2.0 limits frame rate)
  • Power: 1.5–3.5 W; cable-powered

Solves / unlocks

  • Fruit detection + reach distance (single camera replaces vision-stereo pair)
  • 3D crop-canopy mapping for biomass estimation
  • Visual SLAM (RTAB-Map, ORB-SLAM3) for autonomous-rover navigation
  • Pick-and-place robot arms (target acquisition + collision-aware path planning)
  • People/animal proximity detection (safety stop)

Constraints

  • Sunlight defeats structured-light — outdoor performance is poor for Kinect-class sensors; active-stereo (RealSense D455) and passive-stereo (ZED) handle sun better.
  • Range tradeoff — most hobbyist sensors max out at 10 m; long-range outdoor work needs LiDAR.
  • USB 3 cable length — 3 m without active extender; longer runs need fiber-optic USB.
  • Compute weight — point-cloud processing is heavy; pair with Jetson for real-time work.

Source

See also

Auto-generated from this entry’s typed relations: frontmatter, grouped by relation type so the editorial signal isn’t flattened.

  • Member of: [[ingredient]]
  • Combines with: [[nvidia-jetson]] · [[ros2]] · [[opencv]] · [[lidar-rangefinder]] · [[raspberry-pi]]

What links here, and how

Inbound connections from across the wiki, grouped by lens and by relationship. These appear automatically — every entity page declares what it links to, and that data populates here on the targets.

Practical

combines with

  • Acorn rover Acorn's perception stack includes RGB-D for crop-row tracking and obstacle detection
  • LiDAR rangefinder complementary perception: LiDAR for accurate distance, depth-camera for 3D structure
  • NVIDIA Jetson RealSense + Jetson = real-time visual SLAM for in-field navigation
  • OpenCV depth + color frames feed into OpenCV pipelines for fruit detection, weed segmentation
  • Raspberry Pi Intel RealSense D435i has Linux SDK; Pi 5 can run real-time depth processing for navigation
  • ROS 2 ROS2 has mature drivers for RealSense, ZED; standard input to perception nodes
  • YOLO (object detection) YOLO finds objects in RGB; depth channel gives distance to each detection

contains

combines

9 inbound links · 6 outbound