|
Visual Anchor module vs. encoder?
Anonymous |
15 year
|
I have a question concerning the visual anchor module. I know it can track the x axis for use as a visual compass, but what about odometry as a whole? Can the visual anchor module be used in place of encoders, or would encoders still be essential if the robot is going to keep track of its odometry.
I haven't had a chance to try this yet, but I wanted to ask first. Thanks!
Loren John Presley
|
|
 |
John Christian from Norway [25 posts] |
15 year
|
I would think that for rotation visual anchor would be fairly precise (actually better than wheel encoders), but there is no way of using vision for calculating forward or backward motion in general. If the system knows the real size of objects in its vision then you could estimate distance change to the object, hence how far it has travelled, but even this becomes imprecise if the object is not centered in the vision. It would probably have to be a sphere as well so that its equal from all angles. On top of that it would have to be fairly large as well so that motion towards and from it would actually "scale" the object (the smaller the object, the higher resolution you need on your camera to measure anything).
I think a combination of the visual anchor module for rotational data and wheel encoders for forward/backward motion would be the best combination.
|
|
|
|
Anonymous |
15 year
|
John is correct on this. We have tried a different approach by having the camera look at the ground. You essentially have an optical mouse at that point. However, we did not pursue it far enough for a demo but in principal tracking of this kind can be used for odometry assuming you don't get too much parallax (i.e. needs to be a flat textured surface and not too many lighting issues (i.e. cast shadows from the robot itself).
STeven.
|
|
This Thread