RGB-D / depth camera

An RGB-D camera is a camera that captures two things at once for every pixel: its ordinary color (the R, G, and B stand for red, green, blue) and its depth — how far that exact spot is from the lens (the D). So instead of a flat picture, each point in the image also knows its distance, as if every pixel in your photo could whisper how many centimeters away it sits.

It gets the depth without a tape measure using one of a few tricks. Some models project an invisible pattern of infrared dots onto the scene and watch how the pattern bends over near and far surfaces (structured light). Others use two slightly separated cameras and compare the small shift between their views, the way your two eyes judge distance (stereo), or they time how long an infrared flash takes to bounce back (time-of-flight). The result, lined up with the color image, is a colored 3D snapshot.

For robots this is a cheap, compact way to grasp the layout of a room or a tabletop: where the cup is, how tall the box is, how far the wall stands. Compared with LiDAR it is usually cheaper and gives rich color too, but it typically works best indoors and over shorter ranges, since bright sunlight and faraway or shiny objects can confuse its depth reading.