WHAT IS ROBOT VISION?

As the term said itself, robot vision, is a functionality by which a robot will be able to see and perform visual tasks. The whole science behind robot vision is always been an interesting subject in robotics. People started working on robot visions in the 1980s and it is now growing more and more with time, as machine learning is improving. There are different factors working behind how a machine will recognise an object just by vision. 

Let me show you in detail:


MORE ABOUT ROBOT VISION...

Robot vision can be implemented using some basic technologies like signal processing, image processing, machine learning, computer vision and machine vision.

  • Now comes to signal processing, why is that! Because you cannot process any kind of image without processing the electronic signals. Processing signals means cleaning the noises, extracting information or showing the output like anything can be a signal. Some basic types of them are digital, analogue etc. and an image is nothing but a 2D signal or even more.

  • Hence signal processing is important. And if you are looking for the contribution of image processing then here it is. Sensing an object then digitizing it, and after that storing, segmenting and enhancing all are done under digital image processing. So, basically, this one is definitely the important one.

But we have another question! Machine learning! Really!

  • No, people, just don't avoid it, it holds the same importance as previous technologies. Think, you have seen a new animal today, what would you do first? You are going to observe the features of the animal and try to match the pattern with your known animals, right! Hence this is exactly done by machines too. If you want a robot to understand any object by vision, then you have to train them with different patterns and previous datasets. And this task cannot be performed without pattern-matching algorithms and other information-based machine-learning algorithms.

  • A doubt can arise by listening to the next terminology we are going to use, "computer vision", how is it different that image processing? Yes, image processing is ALMOST near to computer vision, ALMOST. Computer vision is generally used to extract information from an image and those are then used by the ML to recognize patterns. While image processing can only be used to store and improve the image.

  • Finally, it's time for machine vision, in this one, a machine uses computer vision and machine learning to understand the visuals as well as guide automated machines for industrial usage like automatic inspection, process control etc. There are some basic types of machine vision available:

  • 1D Vision System 
  • 2D Vision System 
  • Line Scan or Area Scan system 
  • 3D Vision System

After merging all the above-discussed techniques, when we incorporate the kinematics, reference frame calibration, and physical movement of a robot with it, it changes to a full-fledged robot vision. 

SOME APPLICATIONS OF ROBOT VISION...

Robotic Arm Computer Vision, Universal Robot Vision System, Stereo Vision Robot, Welding Robot Vision System, Vision-based Mobile Robot Navigation etc.