0
$\begingroup$

Rosanswers logo

I'm interested in developing a robot (UAV) indoor navigation system basing on a MS Kinect (or Kinect-like, like the PrimeSense dev kit) sensor. I'm specifically interested in corridor following.

I've recently bought a BeagleBoard-xM platform and installed Ubuntu and ROS on it. I'd like to start off without any flying platform and implement the navigation layer first. So here is my plan (and questions):

  1. I have to buy a sensor for VSLAM (ie. Kinect or so): do you have any suggestions on which one should I choose? I unfortunately wasn't able to connect PrimeSense recently, is it even worth it, or should I stick with the original Kinect?
  2. Which stacks should I start playing with? I'm particularly interested in indoor navigation and attitude estimation (I could fuse with IMU based attitude data afterwards).
  3. I would then try to implement a distributed system (Desktop PC + BB-xM) via ROS, with both peers doing a part of calculations. How would you design such a system, what potential risks do you see right away?

Any suggestions welcome, tom.


Originally posted by tom on ROS Answers with karma: 1079 on 2011-02-16

Post score: 2


Original comments

Comment by mmwise on 2011-02-17:
when you see an answer you like, mark it as an accepted answer

$\endgroup$

3 Answers 3

0
$\begingroup$

Rosanswers logo

I have to buy a sensor for VSLAM (ie. Kinect or so): do you have any suggestions on which one should I choose? I unfortunately wasn't able to connect PrimeSense recently, is it even worth it, or should I stick with the original Kinect?

I was also unable to get a Primesense SDK from them directly, however you might want to continue trying as the Primesense device is quite a bit smaller, lighter, and less power hungry than a Kinect (which certainly has to be a good thing for a UAV).


Originally posted by fergs with karma: 13902 on 2011-02-16

This answer was ACCEPTED on the original site

Post score: 3


Original comments

Comment by fergs on 2011-02-18:
I only got one by winning one in the ROS 3D Contest

Comment by tom on 2011-02-17:
Thanks fergs, and is there a way to get a Primesense SDK indirectly anyhow?

$\endgroup$
0
$\begingroup$

Rosanswers logo

  1. You could start with the Kinect and have a look what Patrick has done in the 3D contest. altitude control However as your onboard resources are limited (unless you're able to compute on the pointclouds from the internal DSP, I'm very interested in this, but I have no time) a combination of a regular camera and ultrasonic height measurement will do just fine I guess. We'll probably have some thesis students working on this also.

  2. Look at this question + ansers: monocular VSLAM

  3. Look at this question: kinect-data-via-router-to-host-pc and my answer: answer


Originally posted by KoenBuys with karma: 2314 on 2011-03-02

This answer was NOT ACCEPTED on the original site

Post score: 1


Original comments

Comment by tom on 2011-03-02:
Using DSP would be great, I don't know if it's possible on xM with Ubuntu though, see the information about a bug in xM here: http://elinux.org/BeagleBoardUbuntu#DSP . I don't know how to understand this, whether a hardware or a software error is causing problems. I'll read more when I have time...

$\endgroup$
0
$\begingroup$

Rosanswers logo

  1. ASUS are supposed to be releasing the Xtion PRO Developer Kit, an Primesense SDK based device. It may be easier to obtain than the Primesense SDK. Details are sketchy for a Q2 2011 release. Note that some photos show it without the color sensor (ie. depth only).

  2. Have you tried playing with the RGBD-6D-SLAM entry in the recent OpenNI competition?

  3. Given bandwidth limitations, you may want to decimate the point cloud. This tutorial may be a good starting point.


Originally posted by rreid with karma: 124 on 2011-03-21

This answer was NOT ACCEPTED on the original site

Post score: 1

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.