Networked RR Servers
17 year
I have a mobile robot running Roborealm. Unfortunately the remote mobile agent's CPU  is not fast enough to process images. I would like to have another computer running Roborealm , read the streaming images from the mobile agent's Roborealm  and process the images.  I have used the RR python API to retrieve images from the remote agent,however after several minutes the link crashes with the message "would have to block".

My question is as follows: Is there a way to have a version of Roborealm access the images from a version of Roborealm running on a remote machine?

Any Help would be appreciated.

Thanks G Byron
Anonymous 17 year
Yes, have a look at Distributor Client and Distributor Server in the docs. Those modules are mean for distributed processing and were used on our bucketbot a couple years ago to increase its computational power.


Let us know if you need any more detail other than what's in the docs.

Face Tracking
17 year
Thanks STeven. I just found that service. It seems that all one has to do is read the darn manual.  The Distributor/Client function does exacly want I wanted to do, without the blocking issue.

I truly appreciate the work and help you and your team provides this ragged  community.

I have another question. If the answer is in the text then a pointer would be appreciated. I have been looking into OpenCV ( with the python API). Is there function or RR Algorithm that implements a face tracking /localization function?

Again any pointers would be greatly appreciated.

Thanks GByron
Anonymous 17 year

Thanks for the post!

In terms of a face detection module, not currently. We're working on a face detection module but are not satisfied yet with the results in terms of accuracy and speed. Look for that to come in a couple months (it is still very much in the research stage.) Sorry for the wait!

Anonymous 16 year

Just wondering about the progress towards introducing a face detection module for RR?  For the past few months I have been working with the LEAF project group on my own leaf robot.  They use a open CV vision system for face detection, using eigens.  It works sometimes.  With the new API for RR to lisp, I am sure the group would like to see a better working face recognition system.  

Really, just wanted an update.

Anonymous 16 year
Hi Bob,

Yes, we had been doing some research into face detection not unlike what OpenCV does (note that the eigen value technique used in Leaf is for face recognition which is a process after face detection). We thought that face detection would be an easy process as OpenCV appeared to be doing it quite reliably and others in the research literature also claimed to be quite successful in face detection (we will leave out any discussion of face recognition as detection is normally a pre-requisite for recognition) but after testing some of the techniques out we were somewhat disappointed with their performance (you have evidently reached a similar conclusion).

But it turns out that reliable face detection is a very difficult problem when the background is not determined, when the faces are not necessarily frontal or when the faces are not oriented at 90 degrees and you want to be able to detect faces at around 20fps ... which were some of the requirements we were attempting to forgo. So, unfortunately, research and experimentation continues but there is no estimated time for completion and with other more basic robotic issues like navigation taking precedent it is unlikely that the face detection module will be out anytime this year. This does not mean that a face detection and recognition module will not appear for RoboRealm but does mean that it will not be anytime in the next 4 months.

Thanks for your query!
Mano from Netherlands  [27 posts] 13 year
Any updates on the face recognition?
EDV  [328 posts] 13 year
Face recognition by AVM algorithm: http://www.youtube.com/watch?v=oRpXBOX5rxY

See more at: http://www.roborealm.com/help/AVM_Navigator.php

This forum thread has been closed due to inactivity (more than 4 months) or number of replies (more than 50 messages). Please start a New Post and enter a new forum thread with the appropriate title.

 New Post   Forum Index