ccapriotti's profile - activity

Hello all.

Trying to get started with ROS, but not always having my development hardware with me, I thought of using a virtual machine.

Turns out I found at least two option of Ubuntu and some other distro with ROS already installed. Great, right ?

In theory they are Open Virutal Format, BUT, trying to deploy them simply does not work. It seems that the schema reading the internal XML file cannot find the expected information.

I've tried this with ESXI 4.0, 4.1 and 5.5, with pretty much the same result, and I am assuming those images must run only under QEMU or similar.

Does anyone know of a VMware-compatible image I could use ?

Thank you.

@MuraliKrishnan you are right. Ii was going over my notes and I realize not only it lacks details, but also that it is confusing. I mixed parted instruction with the later use I made of gparted itself. Give me a few days. I'll repeat the operation and change my post.

This has already been discussed on other topics. Turns out the cost (money and man hours) to keep this kind of images AND maintain binaries+repositories is high.

Not practical in the end. Community goals do not lean that way. (which is a nice way to say that those platforms are not that popular).

I know... the name does not inspire great confidence, but it is not my fault.

There really is a RPI-like (or BBB-like if you will) board that is called Banana Pi. In my opinion, its advantage, besides the Dual A7 processor, is the connection to a SATA device.

Can you imagine one of those with an SSD ?

Well, to the question: has anyone done ( or heard of) any ROS/robot development with one of those ?

Now, yours is a slightly fresher approach to the problem; So far I haven't seen anyone discussing Kinect2 on Windows with ROS.

You will see this answer here , from @BennyRe, stating that, indeed, there is no current support for Kinect2 on ROS, that some people is indeed capturing data on Windows and using on Linux, but all on Windows... As I said, nobody really discussed for a while.

The limitation seems to be on OpenNI, which usually handles Kinect data; They still haven't figured out how to make it "talk" to ROS. Protocol. So, my educated guess would be that THIS needs to be developed, platform-independent. Not quite sure if this is what you are proposing though.

Also, are you accessing the internet via a proxy ? Anything particular about your network ?

Hi. Just for a matter of organization, when getting back with more details, try to edit your original post, instead of "answering it". Now, for your problem, are you saying that those failed links work on a web browser ON UBUNTU itself, or is it a different machine ? This is very important.

On top of Hendrix's suggestion (meaning after you do an apt-get update) you might also want to try this:

apt-get --download-only

in order to force Ubuntu to download all of your stuff first; No install will take place. If the operation fails here, then, yes, it is a repository error. If it succeeds, move to the regular apt-get install and check if the error happens again.

Honestly, Hydro on Ubuntu 12.04 is VERY vanilla. Shouldn't fail at all.

Just in case, post you entire apt-get command, so we know what you are installing.

Looks like someone had a similar problem in the past (April) and a workaround was offered:

http://answers.ros.org/question/11438...

You should also look at your router's configuration.

Some of those routers have a QoS - in you case seems to be QSS or something like that - and those CLAIM to prioritize multimedia, and other sort of traffic. You would expect that feature to prioritize your video stream, right ? Yeah...

I have a similar model and, let's face it, it is a cheap router, with simple resources. Very good for home use, but now we are talking robots, and high speed data.

Just turn that off. It messes with your bandwidth in ways you cannot imagine, not really making it better.

So far there does not seem to exist a driver for v2 of Kinect. According to other posts, people using v2 are doing so "mapping" a Windows-installed Kinect2.

Hmmm. Tough one.

Troubleshooting from the system point of view, a "broken pipe" is very typical of network, but could be that one of the programs died, either the "talker" or the "listener".

Any PID changes suggesting if any of them died ?

Instead of using the name of the computer (supposing you are), have you tried:

use 127.0.0.1

cheat your hosts file to always point the machine to its IP

have only one IP

check system logs for errors.

@DEvnoW, I beg to disagree. @hsu explained a situation that was temporary two years ago. It is not your fault nor it is hsu's but two years w/o any changes on a situation like this, can hardly be considered an "answer". It was a reply, yes, but neither answer, nor solution.

Well, yeah, for ubuntu users and x86 architecture, it may not be a limitation, and, since the main devices here, on this question, are notebook-driven, your point is valid. Still, in a ROS-related discussion, yes, it becomes a limiting factor for several other platforms and "quick" implementation.

Well, "missing" is not the word, but important change: OpenCV - if I remember well - is no longer part of the deal (cannot really say "distribution"), having to be installed separately. This might affect your plans for the rovers.

Geeezzz, I meant Hydro and not Groovy. Changing the main post for clarity, and making my apologies public.

Hello, Jonathan.

I was going over a few of your previous posts, and your project starts...well, NOW. July.

There are still a few things pending on Indigo, as far as i could see, and there is a small chance it could jam your project, just because it is not solved in time.

Solution here is not the one my heart would love (bleeding edge latest, because new is better, right ? ) but the sensible one instead: rock solid performance, on stable, well known platform.

I would go for 12.04 and Hydro. At least you have a lot of people with experience to support you here.

Besides, U 14.04, with the the graphical-rich GUI, is using more resources than previous versions. You might want to consider Lubuntu, or other light-weigh variations (of 12.04). Or even run w/o GUI at all.

On the flip side, how long will this project last ? If it will go on and on for years, then it might be worth taking your chances with 14.04 and indigo, but always keep in mind: new software usually takes more resources, and you seem to have limited resources. But, then again, it is a good way to simulate a mission to mars. (What a dream project).

Note: for the implementation of network latency and bandwidth, I would suggest a FreeBSD server with the "pipe" implementation on ipfw, the native firewall, per IP address (of the bots/rovers). Fixed, and locked to the MAC address, just in case you have some little hacker among your students ;)

@DrBot, you mentioned running Ubuntu on a Raspberry PI ? AKAIK, ubuntu did not support (12.04 at least) ARM6. Have I missed something ?

Well what do you know... Same question a few minutes earlier.

http://answers.ros.org/question/18664...

It helps to know what hardware/platform we are talking about. x86 installs and ARM have different statuses.

Two points: the file .bashrc~ is NOT executed. the TILD in the end means it is a temp file generated by a text editor.Second, on an earlier post you stated that you wanted "to run on the local machine...". Clearly it is not the case if you have the robot and a workstation, and need to exchange data.

Dambrosio, hello. I do not have an answer for you, but this was one of my objectives; to test groovy on 14,04. It is not a matter of working or not, but rather of being supported or not. So far, it is an unsupported platform, but go for it. If you have the means, test on a VM with 14.04.

I think this needs some clarification. After your first run, when you are sure that the problem is present, can you run cat ~/.bashrc and post the results ?

Alex, what our crowd here means is the following:

As you probably already know, ROS is not an operating system per se (contrary to what the name leads you to believe), but a giant library of applications that, integrated, will be able to control a robot.

Let's call ROS here, in this context, a layer.

Now that layer depends on another, which is the operating system. In most cases, ROS runs on some variation of unix, typically linux, and on unix pretty much everything is a file, or treated as a file.

More to the point, when instructions tell you to create a file like that, they actually mean : open your preferred text editor, write your code, and save it with name X".

Depending on which variation of unix you have, different text editors will be available.

If you are using a terminal (text based commands, which I think you are), try

nano src/talker.cpp

It is a simple text editor with some hints of its usage at the bottom of the screen.

Or, if you are using a GUI, AND a terminal on it, you can try

gedit src/talker.cpp

which will probably bring up a nice graphical text editor, which is pretty much user friendly.

As you progress in code complexity, other editors, or even programming environments may become more suitable, but for simple stuff, stick to the basics.

Now, if life finally starts to make sense with all of this blah blah of mine, I would suggest that you take some time and learn a bit more about typical unix commands, most likely "bash" commands. That will come in handy.

Have fun !

But if this is the scenario, I'd go for two motorized wheels and one free wheel, holonomic. IF this is an option, of course. We would need from details from @atanea.

Actually, I did the opposite: I read "non-holomoni" (and had to look it up) and forgot about the non-differential. It gets confusing, but, if you stop to think about it, it is simpler from a mechanical point of view: two motorized driving wheels and two steering wheels. (cont...)

David, I am not sure if I am missing the point here, but navigation for Ackermann-type steering is far from optimal, AFAIK. The planned route looks good, even including k-curves, according to documents I found, but executing the plan, K-curves are solemnly ignored. Tough for car-like robots.

@TT: can you post your system specifics, like Linux distro, hardware platform, etc ? Since this is solved, we might as well document it; Might can help other souls going through the same calvary. I will vote you answer right, since you with your current karma points may not be able to do so.

Unless there is a ROS limitation to the number of serial ports to be connected, I do not foresee any issues. Programming the Arduinos will be simpler, and your students will be able to focus on learning programming, physical interface and electronic, and ROS will be in the BG, overseeing all.

A few generic points to help troubleshooting. Compare top results when the bug happens and during normal operation. See if there is a process taking more time. Also, it might be that some process went to sleep mode, like a power save feature.

Just to add my 2 cents, regarding Arduino: from a student POV, it is great, because you can learn how sensors work (digital are either up or down, analog ranging from 0 to 1023), voltage divider, and all that good stuff.