Terminologies used in RTAB

I could not find the explanation for the following terminologies mentioned in the course. It would have been nice if these topics were explained in the course itself. The topics are:

  • What’s a bag of words approach?
  • What’s a loop closure detector?
  • What’s an ICP?

Hello @Joseph1001 ,

  1. “bag of words” is a machine learning model which looks for patterns in image features that can be used to classify various images as belonging to one location: Bag-of-words model in computer vision - Wikipedia

  2. A loop closure detector is a method that detects when the robot has returned to a past location (which has been visited earlier).

  3. ICP stands for Iterative Closest Point. It’s an algorithm used to minimize the difference between two clouds of points: Iterative closest point - Wikipedia

I’ve linked Wikipedia sources, but you can look for many other resources if you’d like to go deeper into some of these.

@albertoezquerro , Thanks for replying

It would be great if “Machine learning” was mentioned as a prerequisite so that users would be better prepared with the content.

Hello @Joseph1001 ,

Actually, I’m not quite sure about this. I think it should be kept separate. Machine Learning is a HUGE subject that can have hundreds of applications. And it’s not really necessary to know about machine learning in order to use the RTAB package (as an average user). I think we should, however, clarify these concepts you mentioned in the post so that students who are interested can dive into them in case they want to learn the insights of the process.

Hi @albertoezquerro ,
I agree with you, i would be great if the general concept of these advanced topics was explained before hand, like say how “bag of words” works (nothing on detail, but some pictures and brief working methodology)

Thanks for looking into this so quick.