Why Collaborative Robots Need to See

Collaborative robotics - robots working alongside their human counterparts with minimal safety structures or risks - has long been a focus of robotics engineering. Machine vision helps make this possible. New frontiers in labor automation are opened when machine vision is coupled with cutting-edge manufacturing techniques, giving robots the power to interact with the world using greater precision than ever.

The Jump from 2D Vision to 3D Vision

When it comes to robots, especially in manufacturing, 2D vision has been around for quite a while. 2D vision systems allow robots to detect movement, localize parts on a conveyor belt, and perform other functions. As impressive as these systems are, the scope of their use is somewhat limited: 2D systems are excellent for basic inspection, but must be coupled with 3D systems for precision in most manufacturing and logistics applications.

One example of early 3D imaging systems is infrared sensors, often used to help verify dimensions of palletized products prior to shipping. A combination of 2D and 3D sensors allows robots to go beyond product inspection to more sophisticated product handling tasks. The ability of robots to integrate wavelengths invisible to the human eye also makes them ideal for specialized environments involving radioactive or other hazardous materials.

Robust 3D vision provides a number of great advantages over pure 2D systems:

  • The ability to record and measure depth, a crucial input in logistics.
  • More precise handling of components in high-speed manufacturing.
  • Easier processing, troubleshooting, and quality control by engineers.
  • Improved management of components at small manufacturing scales.
  • Enhanced prevention of collisions (working with on-board sensors).

The Question of Calibration

Many foundational machine vision systems use pre-calibrated cameras that take a cue from human vision in developing 3D images from multiple 2D perspectives. Passive imaging uses two cameras mounted on the robot, while motion stereo describes a single-camera implementation that derives its vision from images taken from two or more locations.

Up until recently, one of the biggest issues with these systems was calibration. Robots left the factory in a pre-calibrated state with all on-board equipment in its right relationship and its software designed to match. Simply jostling a device would cause it to go out of alignment, requiring a specialist to make an on-site visit to restore service.

In coming years, the top benefit cutting-edge 3D machine vision systems bring to many enterprises may be their ability to self-correct and re-calibrate. Just as you calibrate receivers or cameras by aiming them, advances in 3D imaging technology promise a future where devices can use their own triangulation techniques to calibrate with only non-technical assistance.

For Collaborative Robotics, the Future of Machine Vision is Now

To achieve the highest standards of productivity, efficiency, and safety, collaborative robots need to “see.” New, highly-acute vision systems will be instrumental to maximizing the power of collaborative robotics. As time goes on, it will also be central to producing the kind of adaptive robots who can respond effectively to changes in their own working environment.