In the world of machine vision, we talk about X and Y for locating an object in 2D but we can get Z with stereo cameras. So, as we know, cameras in a stereo vision configuration can be used for depth sensing or height measurement.
If we can add a measurement in the Z axis, we now have a 3D system which opens many additional possibilities over and above the benefit of depth sensing (distance from cameras to target). See a typical Scorpion 3D Vision application for a great example.
For a real time depth sensing application (autonomous vehicles are an obvious benefactor) we also need synchronised images from the two sensors - we know from talking to our customers that this is one of the challenges even before they create the code that extracts the depth measurement.
Depth Sensing in Real Time
For simple depth sensing using machine vision, it can be expensive and time consuming setting up dual cameras, triggering both cameras simultaneously and processing the image data on the host computer.
Also, perhaps this is something you would normally only expect to see on an Intel or Nvidia powered computer platform.
However, Arducam, the innovative embedded camera manufacturer have designed a fully synchronised stereo vision HAT (Hardware on Top) for the Raspberry Pi.
There are currently two versions available with 5mp or 8mp sensors. Both camera systems have integrated autofocus lenses so the cost is limited to just the Stereo package and the Raspberry Pi. No other hardware required.
Typical application areas are obviously flying drones, subsea ROVs and autonomous land based vehicles.
Watch the video to see a demonstration.