I feel like poking the bear... Did you achieve the impossimpible?

It's all gone a little quiet and no news on this project. I'm actually very interested in knowing if a work around to the IP conflict problem and the single Movement Panel issues were found. Have you managed to do it?

RICH havent had the time with my beacon project,my wifi design and then friday leaving on my 3 week work trip

BUT with me i dont give up,and will try my solution,it may not work,but without trying i never will know

Hopping it can be done

Hello Everyone,

This is my first post on EZRobot. I believe I have some relevant information to contribute to this thread.

First, I will say it is possible to control multiple robots, but I don't think robotmaker is on the right path, At least from a quick read. But, I applaud his determination. If you do get it to work using your method, I will be the first to drool.

I say it is possible because I am working on the same project. In fact this is a year old project that I inherited as a class project from a friend who interned at a space agency, which will remain unnamed. He was able to control four robots simultaneously with one pc. Thats right, ONE PC. of course, he had to have a separate WIFI dongle for each robot.

Rich is correct that their would be a IP conflict, but only if that specific IP is called.

The key was to bind each wifi card. What this means is to create an Object and bind that object to a specific WIFI card. My friend accomplished this by first creating an API for the Rover using Mathworks MATLAB. Each Rover has a MATLAB Object created for it. Then that object is binded to the wifi card using Java code.

Now we are trying to figure out how to do this with one wifi card. We want to create a swarm platform, and having to purchase a separate wifi dongle for each Rover is impractical. But we might have no choice.

We are currently migrating our code to ROS using Python and C++.

Don't know if this same method could be applied to EZROBOT as I am unfamiliar with the software.

I have been working in the lab all day, time to hit the bars.

Good Luck guys!

-Berkeley

Thanks for the input @berkeley110.

However, there are a lot of problems with this entire idea. Even after the networking issues are solved it's still riddled with problem that have been flagged up so I wont repeat myself.

As far as I know, a wireless NIC can only connect to one network at any one time. It would be impossible to connect more than one rover to one WiFi card. It may be possible if you could somehow reverse how it works and have the rovers connect to the PC rather than the PC connect to the rover.

Why is one NIC per rover impractical? Wifi NICs are a dime a dozen, and a powered USB hub solves any problems of low USB port counts. However, all of that is pointless if the IP of the Rovers is unable to be changed, which without custom firmware that either doesn't exist or is very well hidden isn't possible.

But even after all of that, even if you achieve connection to multiple rovers with one PC, there is still the problem of ARC being hardcoded to connect to 192.168.1.100:80 for the Rover Movement Panel and the problem of only being able to use one Movement Panel per instance of ARC.

Please keep us updated on your progress though, it sounds like you are on the right tracks (although I remain to believe even those tracks lead to a dead end).

Rich

Hey Rich, he says it works. It's currently 4:30 and I'm dead tired, did you just tell him that he was wrong? I don't believe that he would just lie about it working. Unless we're talking about how it's going to work on the EZ-B, then I'll just go to sleep.

Erreul, you misunderstand. Please re-read.

If it isn't working with ARC or the EZ-SDK then it isn't working. So yes, it is exactly right that we are (at least I am) talking about how it's going to work on in ARC. Have some sleep

Thats same idea the BERKELEY110 is doing is same idea i have ,using 2 WIFI'S But also trying another idea too.

Right now can test my theory till i get back from my work trip in china,but will be my first project once i get back Already have my other USB WIFI dongle in

Hello Rich

Yes, after further research and a discussion with a professor of mine (A Google Engineer, too! so cool!) The only way to accomplish my goal with one NIC would be to create my own protocol on top of UDP, which would ignore IP add. in favor of using the Mac Address. He told me this would be a huge time consuming project and that it would be more practical to go the route of hacking the robot instead. But, for the focus of my project and better focus my time, I should proceed with multiple NIC.

The reason multiple NIC are impractical has to do with scale. Sure, if I have two or three robots, then there is no big deal with a couple of dongles. But my project needs to be able to scale to huge teams of robots who operate as individuals. We are talking about 10, 20, maybe even fifty robots at a time. If I can't easily scale the project in the future, then it would not work great for future research. But, nonetheless I will carry on with my puny three...

I hope that clears up some questions. I will ask my friend if he has any video of the four robots working together. If not, I will ask him if we can run the old setup for some video. Wish I could release the old API he created so you guys could play with it, but I don't think he really has the rights to it.

I plan to make any code I create on my own time available to the pubic. Most likely using GPL or something similar.

Do any of you guys have experience creating and maintaining a blog? Any recommendations? I would like to start keeping an online diary of projects to help others. Thanks for any advice.

-Berkeley

P.S. Apparently there some windows software to create a virtual NIC to connect to two signals using one actual NIC in order to increase download speeds. It came up in a google search. But my project goes in a different direction and I will not bother trying to get that to work.