Rule 4.4 Remote Kill (Advisory Only)

[m2xt]

Eddy, I have seen a number of bots out of control due to a) speedo failing with a shorted FET driver (quite common on NCC70s) and b) due to a CPU crash - if you say watchdog in reply then think again! In both of these cases attempting to turn off the driver enable will have no effect. I still believe that a remote kill that switches off the FETs is fine in most cases but is not as good as a electro-mechanical power interrupt but far better than nothing. I will never run a heavy without a power relay circuit, if nothing else, it has saved the embarressment of a runaway bot following a Vantec crash.

Paul

[fwsorg]

Thank you Paul. That would suggest that the rule may need perhaps some re-wording, as it is not too difficult to take a seperate channel of the TX/RX which should bring the robots failsafe devices to the pre-set off or zero position via a switch on the TX. as meaning that making the controllers failsafe as they normally would is acceptable. As you have described above, that seems to frequently be insufficient. Perhaps in the next revision of the rules (whenever that may be- I seem to remeber Andy Kane mentioning that due to insurance they are either fixed now, or soon to be fixed, for a set period) a note be added on what is deemed acceptable- i.e. that relying on a uProcessor of FET driver is not really suffiecient, or at least not the prefered method? I have no problem with conforming to either of these, but will from now one favour the relay approach.

Thanks,

Eddy

[jim]

Good to know Im in the clear about the CO2 tanks. No need to apologise, Ed - if you have concerns you have to raise them even if you dont know everything about the field in question (who does?)

Whatever remote kill system you put in place can of course fail but if you make your weapon or drive reliant on more than one system (speed controller plus relay for example) and you keep them as separate as possible you drastically reduce the risk of a runaway robot.

I will see if Maplin have any of the 180A relays tomorrow for my robots. I suspect that space rather than weight will be a problem for many existing robots though.

[fluppet]

This is where my not-yet-roboteer paranoia comes in. Id been wondering about failsafes (for heavyweights). Although I wouldnt want the failsafe to fire when not needed, Id want the fail safe to be *really* safe.

The old Phoenix mainframe in Cambridge used to have a shutdown which fired a glass bolt through the power line (we hypothesised it was for use if it ever became self-aware). Id kind of been thinking along those lines for the robot - chomp a (replaceable) wire as it leaves a battery terminal with a spring-loaded nonconductive (ceramic?) blade of some kind, with a small solenoid holding it in place. Strikes me as a simple and pretty reliable system, and probably not too heavy.

I like the infra-red backup idea, by the way.

Would I be vastly over-engineering this?

--
Fluppet

[bugs]

Euhm...
YES!

[fluppet]

Just so long as I know. :-)

--
Fluppet

[fluppet]

Okay; Im just particularly allergic to pain when it comes to being attacked by 100kg robots. :-)

If someones wandering around with random radio signals (as at the Robot Rumble 2003 when some kids were playing with an r/c car outside) or if theres a chance my electronics might go haywire - always a possibility, if someones stuck an axe through them, or if theyre designed by someone as incompetent as me - I like the idea of being able to kill the thing. Being able to do it in such a manner that it cant get switched back on again would be favourite.

Of course it does mean Ill have to make damn sure that nothing activates it when I dont want it to, but thats a small price to pay. Im made nervous enough by the idea of sticking my arm through some railings to stick the link in - trapped limbs and snapping noises haunt my thoughts, and in the past Ive proposed the idea of a cattle crush to hold robots still before entering the arena. Trying to turn off a robot which is going King B or doing a Behemoth (vs Mute) isnt my idea of fun, if the control system is already known to be flakey. I have an innate concern that relays might weld themselves in nutty robot mode, and for as long as I might consider things like flippers which automatically fire when somethings on them, I want to be sure my failsafe wont get stuck.

I trust (relatively) everyone else to be competent with their failsafes - but since Ill be the one with my hands on the removable link, I reserve the right to be very paranoid with my own! (Its just nice to know *how* paranoid!)

Anyhoo, back to people with less extreme solid-state systems (which Ill probably use as well - no need to spend more time than necessary replacing wires!)

--
Fluppet

[wenman]

Ian as Paul says it is quite possible for speedos to fail in a state where they just run even without a signal. In this case the failsafes will not stop a run away bot.

I actually had this happen on my weapons speed controller once, luckily it wasnt connected to my disk at the time. Thats why Im intending to use a mechanical relay this year.

Im using the Ian Watts interface as well this has two weapons channels & I intend to use one to control a mechanical contactor. Now of course someones going to argue that I shouldnt use the mixer to control the contactor in case my mixer fails .

Richard Wenman
Team Mayhem