802.11

Not sure if this is the place for this, but...

I cant be the first person to have wondered this. Would there be any mileage in allowing 802.11b/g (WiFi, except to pedants) as a transceiver frequency band? It implies a degree of intelligence on the part of both ends of the link, but if theres a fair bit of electronics handling a complex robot anyway then a TCP/IP stack isnt so hard to add.

Obviously Im assuming that the device is going to do some decent error recovery, use some secure protocol to avoid accidental interference, have the obvious failsafes, and so on. Nonetheless, stuffing a cheap pcmcia card in a machine seems less hassle, once you have a level of intelligence on board, than playing with 40MHz radio modems. Note that I have no delusions about burying a laptop on board and expecting it to survive, although nano-ITX and a lot of padding might be interesting.

Even in combat conditions Id expect relatively decent bandwidth, and there wouldnt be the issues with crystal swapping. Bluetooth doesnt have the range, but 802.11 does. 802.11a has fewer interference issues, but the 2.4GHz variants are a bit more common.

Has anyone tried it? I can believe there are latency problems if theres a lot of interference (it might be helpful to try to get the audience to turn their laptops off!) but I dont see many other downsides.

Aside from protecting the hardware, the small motherboards are cheap, the cards are cheap, and you can drive it from a laptop - all of which would save me playing games with electronics which Im more likely to mis-use, and which is more expensive.

Anyway, just wondering if there was a reason not to allow it - or, if event organisers are happy with it, to explicitly allow them in the rules. If so it might be wise to request a bandwidth limit (otherwise the electrical jamming rule could be infringed by someone sending video back from a camera on the robot) but, safety checks aside, I dont think it would influence much else.

I dont need it for any robot Ill be building just yet, but it would be nice to know its out there in the future.

--
Fluppet

Youre right, you are not the first one to wonder about it. Just as i wondered if the IFI Isaac 16 Control System are allowed in the FRA rules.

I could not find a specific ruling for them, and therefore concluded that they are not allowed.

It might be a suggestion to add them or at least discuss adding these into the rules.

For specs on on th IFI Isaac 16 Control System, see: http://www.robotcombat.com/marketplace_ifirobotics.htmlhttp://www.robotcombat.com/marketplace_ifirobotics.html

--
Leo

I cant see any real problems with WiFi to be honest, other than I seem to remember Mentorn not being too happy about data being transmitted back from the data (although this was intended for telemetry AFAIK). It wouldnt be easy, but itd be an interesting project.

I looked into the ISAAC systems some time ago and couldnt see it being an easy task to make it operate on UK frequencies, I think they are intended for 75MHz.

If you put a camera on your robot, stick a GPRS card into your laptop and connect it to the internet then you can effectively control your robot from anywhere on the net! Just think of driving from the comfort of your armchair at home, whilst the rest of your team do all the heavy lifting work! Wouldnt be as fun though would it.

Er... *LAG*.

--
Fluppet

75 Mhz?

quote from IFI:

Notice to BattleBots„ Competitors: The use of these modems for the purpose of BattleBots„ is restricted unless used with the Isaac32 or Isacc16 Control System by IFI Robotics. This agreement between BattleBots and IFI Robotics restricts the use of the 902-905 MHz, and 925-928 MHz portion of the ISM Band to the IFI Robotics systems exclusively.

I think those frequencies are used by GSM mobile phones in Europe.

Either way, it aint a legal band to use over here.

Dave, theres already a guy here in Sweden who have done that!

http://www.fightingrobots.co.uk/discus/messages/25/551.html#POST2074http://www.fightingrobots.co.uk/disc....html#POST2074
http://www.fightingrobots.co.uk/discus/messages/26/220.html#POST2075http://www.fightingrobots.co.uk/disc....html#POST2075
http://www.fightingrobots.co.uk/discus/messages/20/461.html#POST2076http://www.fightingrobots.co.uk/disc....html#POST2076
http://www.fightingrobots.co.uk/discus/messages/10/941.html#POST2077http://www.fightingrobots.co.uk/disc....html#POST2077

He says he gets 0.3 seconds latency - might not sound like much but it makes a huge difference when youre in the arena. Serrano has a similar latency problem in its RC system and its like driving an oil tanker.

Presumably the 0.3s latency is dominated by the internet latency and the bandwidth taken up by the video signal (unless its a completely different frequency). For the purposes of running a local connection Id expect latency to be a few milliseconds, especially if the machine isnt tied up with trying to run Windows as well. (Linux isnt all that well suited either, although theres a bit more control and there are pseudo RTOS variants - Im tempted to see if BeOS has any WiFi drivers).

Trying to run a video signal back the other way (especially if something is playing with inter-frame compression, which is inherently laggy) is bound to cause havoc!

A commendable effort (Id seen the posts before), but its practicality or lack thereof isnt indicative of the result of using the same technology locally.

Ignoring for a minute the question of whether its a good idea, is there any technical reason this shouldnt be considered?

--
Fluppet

I didnt meant to disparage the idea, I think its well worth looking into. It could provide many advantages over the current radio control options.

Sorry Jim - I was a bit more defensive than I meant to be because I wanted to make it clear to other readers that two different things were being discussed.

Obviously 802.11 has overheads in comparison to a simpler system, but as you say it does have advantages once youve already got a control system of the appropriate complexity. (Its a sledgehammer to crack a nut, but if youve already got the sledgehammer and youve got a nut to crack...)

--
Fluppet

Gents,

It wouldnt work in an arena environment. Not enough transmitting power and reception sensitivity. You have to know that WiFi only works well within certain parameters ie. in your home...and not inside a steel cage buzzing with signals....hell it doesnt even work properly in some homes...(eg. when the floors/ceilings are made of fortified concrete..full of iron).

But I would like a way to do two way communication with our robots...eg. for telemetry, health status, etc.

cheers,

Frank

Whoops - sorry, Id half written a reply to Frank, then failed to post it.

I have to say Ive never had a reception problem with 802.11b (with the base station wedged behind large metal boxes), although my house is hardly huge. Ill take your word for it that the arena is too noisy, but on paper the range is good enough, and the fail-back to slower modes should degrade gracefully. Id have expected it to kind of cope (if not at full rate), but if youve tried it and it doesnt then Im certainly not going to argue. It would be embarrassing to find a signal doesnt get through the railings at Debenham, say. (Im sure things could be improved with a high power antenna, but allowing this while disallowing boosted 40MHz antennae seems unreasonable).

40MHz wireless modem it is, then (or the equivalent components). With a bidirectional link, obviously. Now to dig out the part numbers someone suggested to me in the past... (just when £20 of 802.11 cards looked promising).

Incidentally, my feeling is that a lot of fights struggle because of reception problems, and that it might be fairer to try to do some things to the arena to improve reception (pipe the antenna to the middle of the ceiling, add radio-absorbing material to the walls, make the arena a Faraday cage so the equipment outside doesnt interfere, that kind of jazz). Perhaps when people have got rich on their class 1 arenas...

--
Fluppet

I have just installed 802.11 g on my computers.
The Adsl hub/WAN point is indoors in a wooden portacabin.
Having just taken the laptop for a walk outdoors, I found it worked well up to about 100 metres BUT only when stood still.
As soon as I started to move when the range was greater than 20 metres, the link light when out but came on again soon after I stopped moving.
That does not sound good for controlling a moving robot.

Thanks for the feedback, Roger! Ive had limited experience (Ive only got a card in one machine, and Ive cunningly misplaced the charger for it...) so Im prepared to believe its a bit flakey. Id expect 802.11b to do a little better than .11g, but not so much so that signs of .11g failing to cope dont bode poorly.

Ill plan to go down the route of hijacking the signal from a normal 40MHz transceiver for now, then.

--
Fluppet

Would like to discuss this with you on sat Andrew if youre coming to the unofficial social.

Eddy

P.S There was a bluetooth controlled ant at the FeatherWeight Smash 2004. Obviously no good for anything larger, but fun nonetheless.

Im starting to wonder about that Eddy.
It was 100% safe and needed a password either side, on the receiver and transmitter, i cant see anything unsafe about it to be honest.

Mr Stu

Its all a question of what happens when the signal drops, Stu. If the machine carries on what it had been doing regardless, then that is too dangerous for anything bigger than ant, possibly featherweight (assuming no Vortex-style discs in use!) Its only safe if it fails safe properly, and I dont know that the Bluetooth standard says anything about that. And Im too lazy to go look it up right now.

LOL!!!

Mr Stu

Failsafe behaviour operates on a layer above the physical networking. The medium you use, whether 40Mhz, 459MHz, IR, bluetooth, 802.11, GSM, or tin cans and string is irrelevant. If any these standards have any built in safety then I wouldnt rely on it. The content of the message is all thats important. For example, a conventional failsafe module looks for pulses of between 1 and 2 milliseconds, 20 milliseconds apart - it doesnt care how they were transmitted.

Some further info, the 802.11 g runs at 54 Mb/s but only up to about 10 mtrs range.
As the signal gets weaker it reduces its data rate from 54 to 48,24,12,6,3,and down to 1 Mb/s
At its max range of about 100 mtrs, its rate is only 1 Mb/s.
These measurements were taken from the Link Monitor on my laptop when I did the original tests.
1Mb/s is plenty fast enough for a robot data link, but not enough for decent video.
But, as I said before, only when the link is static, not moving.

I wish I had a laptop to test it with! I do know, however, that some people have managed to get WiFi signals in a moving aircraft 2500ft above a city, using a commerically available antenna, so perhaps we shouldnt write the idea off just yet.

http://quickwired.com/kallahar/stories/2003-Dec10/warflying.phphttp://quickwired.com/kallahar/stori.../warflying.php

There are universities who have set up 802.11 transmitters with range over 1km (dumb idea - shared bandwidth, anyone?) but I suspect the transmitter strength rules would be broken. Roger - I didnt have video in mind (although I know its been done in series past), but it would simplify the protocols for some telemetry. Theres no way I need 1Mbps, even for my dafter ideas. :-)

Out of interest, what do the house robots use for their cameras (assuming they dont record locally)?

Eddy - should be there, will look forward to chatting to you.

--
Fluppet

They record locally Andrew.

Ah. Thanks, Ed. Oh well, so much for that theory. :-)

What technology did the competitor (I want to say Lightning, but I think Im confused) use back in series three-ish to do the video feedback? Just curious.

--
Fluppet

Jim, there is a lot of difference between picking up a WAP long enough to identify it and getting a continous signal to control a robot.
If you keep losing the signal your system will go into failsafe mode often and your robot will behave like a kangaroo.
The arena metal framework will play havoc with moving 2.4 or 5 Ghz WAP signals.
So will another metal robot moving near yours.
The signals reflecting off moving metal objects will add and cancel out all over the place.


Andrew, It is possible to get 10 Km with 802.11 using high gain directional aerials and standard power but only fixed point to fixed point.

I dont know what the H/Rs use for video links but the BBC have access to many channels for video transmission.

Both 1.3 and 2.4 Ghz 0.1 watt license exempt systems are available for all to use for video.
They have a range of about 400 mtrs using simple omnidirectional aerials. Cost under £100.

I am a radio ham and I use 1.2 and 2.4 Ghz (at 16 and 1 watt) links for my robot cameras, and get more than 30 miles line-of-sight easily.
I am allowed much higher power with the license.

Very true, Roger.

It would be fairly simple to do an experiment - we would need a WiFi enabled palmtop computer like an iPaq which could fit inside a robot chassis, and a WiFi laptop. We could then rig up a program for the iPaq to record data streamed from the laptop during a fight, preferably a fairly relaxed fight so theres less chance of damaging the iPaq. Does anyone have the right equipment to do this?

Roger, did you get my email about the converted 459Mhz Futaba 9C set you had advertised on the For Sale page ?

Ed
http://www.teamstorm.comhttp://www.teamstorm.com

Roger, got your email saying you didnt get mine ! But when I replied to it I simply got a bounce back saying that your plusdsl address is not a valid address !

Have emailed you at the address on your profile.

Ed
http://www.teamstorm.comhttp://www.teamstorm.com

I have tried again with a corrected(?) address.

Jim, you dont have to use a robot, just put your system, including the output interface and servos (if you have got that far) into a pram and push it around a simulated arena.
See if the output or servos behave(s) properly or if it goes into failsafe mode too often.

I was thinking that the motors in a robot would significantly affect the radio performance, what do you think?

Andrew - these articles may be of great use to you as the goalposts may have now been moved with regard to standards, encryption and hacking etc.
http://www.vnunet.com/News/1153321http://www.vnunet.com/News/1153321
http://www.vnunet.com/News/1124574http://www.vnunet.com/News/1124574

Thank you, Karoline - Ill have a read.

George Francis was making noises about experimenting with 802.11; Im hoping hell post a report. I dont really want to spend hundreds on a transceiver when I can get by with a 20 quid CF card... (Im not saying Rogers wrong, since I know hes an expert on these things, but the more experimental evidence the better.)

--
Fluppet

Jim, I dont think there is much chance of any robot drive motor motor affecting 2.4Ghz 802.11 transmissions.

Just to keep you updated, I have changed over to 802.11a at 5Ghz.
The 2.4 Ghz 11.g was interfering badly with my video transmissions on 2.4 Ghz, whatever the channels in use.
On my walkabout test, 11a had the same line-of-sight range as 11g, about 100 mtrs.
11a did not suffer from the link breaking while I was moving as did 11g (see previous post).
This would indicate that 802.11a is a better choice for moving robot control, and essential if you want to also use a video link from the robot.
The 11a link did break when a large bush was in the line-of-sight, 11g did not.
Only simple tests that need to be backed up by others but an indicator as to which system to use.

Interesting. Thank you, Roger. I hope Ill be feeling richer at some point, and start my own experiments. (Im not intending to be a permanent vapourboteer.)

--
Fluppet