PDA

View Full Version : Wifi to Control AX12 Mech??



DC Robotic
07-11-2009, 09:36 AM
I ran across this WiFly GSX 802.11b/g Serial Module - Roving Networks.
http://www.sparkfun.com/commerce/product_info.php?products_id=9290

I was wondering if it would be suitable to control AX12's over wifi? It looks like it can handle the full com speed of the AX12 bus. Plus it has some capacity for I/O.

I'm still a noob with the communications. I would like to keep all the briains for my mech offboard on my laptop running a C# program. At this point I'm trying to limit the number of new disciplines I need to learn. Would prefer not to have to learn all about programming micros (right now) if possible.

What I would like to accomplish is replacing the USB cable from Laptop to USB2Dynamixel on Mech with wireless solution suitable for Robogames. Also to have enough I/O to fire guns or other simple actions. I also want to be sure not to introduce huge latency issues

I would really appreciate comments/suggestions.

Darrell

nagmier
07-11-2009, 01:14 PM
From what I gathered WiFi was relatively reliable at robogames (I wasn't there so someone who was could answer that better) though xbee seems to be the method of choice

darkback2
07-11-2009, 01:40 PM
I used wifi to control my bots and transfer video. Uniquely I have two PC controlled mechs...As in onboard PC controlled over wifi.

Wifi was reliable...but everyone who was getting video over wifi experienced really slow video, and lost packets were common. There was a lot of interference, and I had trouble keeping people off of my network...probably should have used some sort of WEP encryption...but I was way too lazy...

I think lost packets were common with xbee also...

I would not plan on sending all of your servo positions over wifi...lost packet and your robot becomes...well you get it. The other option would be to have your robot be really slow, and incorporate some sort of error checking.

Ok...so the other idea, have a micro on your bot, and control that over wifi or xbee...

The good news is your at the right place. Check around as your project gets fleshed out more about different micros that can do what you need, and how to do it. I would suggest you go with something like a series of poses, and step through them for each of your gates. One thing that I did that eliminates a bunch of problems is to have each gate start at a common point, and have them return to that point when you stop the gate. In that way you won't have a bot that falls over because it is switching between gates mid stream.

Hope this all helps.

DB

DC Robotic
07-11-2009, 03:21 PM
DB,

Thanks for the response. I'm at the point of buliding gaits for my first Quad and then I have another design I'm going to try to see the difference. The Common starting point or neutral position sounds like good advice. I was also considering dynamically shifting the battery position to assist in stability, anyone tried that?

I'm trying to make the best choice in how to proceede with the communications and where to locate the brains. I had originally intended PC on board by gutting one of the small $200 laptops. Now working with the actual quad I'm real concerned about weight of that solution. Then looked at Gumstix, then Xbee and the way Issy was controlled. Don't mind spending the $, just don't want to do it a bunch of times to find the way that works.

lnxfergy
07-11-2009, 05:17 PM
I'm trying to make the best choice in how to proceede with the communications and where to locate the brains. I had originally intended PC on board by gutting one of the small $200 laptops. Now working with the actual quad I'm real concerned about weight of that solution. Then looked at Gumstix, then Xbee and the way Issy was controlled. Don't mind spending the $, just don't want to do it a bunch of times to find the way that works.

The Gumstix are really just high-end embedded devices -- unless you regularly compile Linux kernels from scratch or write device drivers, you probably want to avoid it, it's really not a PC. You'd be far better off with a small AVR or PIC microcontroller.

Issy's software is currently being rewritten as an Arduino library, ask anybody here, Arduino is pretty much the easiest to use real microcontroller out there. The controller board I was using in Issy will also be available shortly (if you are interested in getting a beta version, PM me as those should be ready end of this upcoming week).

-Fergs

Rbotguy
07-11-2009, 10:54 PM
I have the older version of the Wifly, and it seems to work as described, but the documentation is horrible so it took me the better part of a day to get it running. I still haven't managed to get the "update over ftp" feature working. That being said, as long as 3.3v isn't a problem, I don't think you can beat it for $69.

-Rbotguy

JonHylands
10-24-2009, 01:10 AM
I've been catching up on some old threads, and noticed this one. See my "Bioloid Boards Again" thread:

http://forums.trossenrobotics.com/showthread.php?t=3646

- Jon

Adrenalynn
10-24-2009, 04:28 AM
You really do need to check out the arbotiX too, Jon. Fergs has done a stellar job with his open-source libraries and support code. I bet you would find that you could do magic riding on his work, and I can't imagine a better synergy than you and Fergs banging out code for the arbotiX.

JonHylands
10-24-2009, 09:26 AM
Yeah, I've seen it. Of course, you can't do full speed wireless from a PC, which is what I'm trying to accomplish with my wifi board, but it looks pretty sweet nevertheless.

- Jon