dac@weed.frc.ri.cmu.edu (Daniel Christian) (05/25/90)
Be careful about switching to optical means in a radioactive envirionment. Most optical sensors are just as sensitive to radiation as light. Many device (e.g. optical encoders, opto isolators, CCD cameras) become saturated by radiation. This is one of the reasons the rad hard systems tend to be very simple and often are tethered. The robots used in TMI were all tethered. The robots used at Cherenoble (sp?) were mostly simple radio control systems. NOTHING is easy in radioactive environments... Dan Christion Field Robotics Center Carnegie Mellon University dac@ri.cmu.edu
nagle@well.sf.ca.us (John Nagle) (05/25/90)
In article <$B^$N*%@ads.com> pkahn@ADS.COM (Philip Kahn) writes: >In article <9352@pt.cs.cmu.edu> gerry@cive.ri.cmu.edu (Gerry Roston) writes: >>Here's a problem we are all probably familiar with: how >>to communicate with a mobile robbot without using a tether. >An easy and relatively cheap kluge is to use cordless phones and >standard modems (digital data only). Low bandwith and need to >compress and transfer images slowly is the user's responsibility... Not legal. Cordless phones are type-approved, and user mods are prohibited. See 47 CFR 15.235. It's possible to use the type-approved RF modules which form part of some R/C gear to build a two-way system. I've done this, and it's painful to get it to work. I don't recommend it. The receivers are designed for the pulse train waveform that R/C systems use, and don't work well with other waveforms. Even the high-end approach of getting FM TV microwave gear doesn't work all that well indoors when the transmitter is moving. The reflection nulls get you. A diversity receiver is required. But the way to go is probably to look into wireless LANs, of which there are now several on the market. Any info would be appreciated. John Nagle KA2XXF (experimental, not ham)
khoult@bbn.com (Kent Hoult) (05/25/90)
A few years ago there was a terminal interconnect that used LED "Floodlights" on the ceiling, and smaller ones on the terminals. The floodlights consisted of about a dozen hefty LED's in a clump aimed in various directions. There was also a phototransistor in the clump to receive. I think the idea was that you didn't even have to have direct line of sight, white walls would reflect enough light to keep the link going. If your robot is going to stay in one room, then something like this might work. (PS - I seem to recall this thing was used for IBM 3270 terminals at around 1.5 MBit/sec) Kent Hoult TEL: (617) 873-4385 ARPA: khoult@bbn.com
esrgf@warwick.ac.uk (Mr M Stevens) (06/08/90)
In article <24493@unix.cis.pitt.edu> fmgst@unix.cis.pitt.edu (Filip Gieszczykiewicz) writes:
Greetings. What do you people think of using sound waves
as a transmission medium. I think that anything below
10 KHz would have problems with background noise (pumps,
60 Hz, etc.) but anything above 40-50 KHz should work
just fine. Does anyone see any problems with this?
I've thought of using sound waves for robotic applications a few times
but so far haven't done too much about it. The main problem with
anything <20K is that it going to drive every person/dog/cat in the
vecinity crazy.
Higher frequencies (ultrasonics) are of course used a great deal for
obstacle detection / range finders. We have some very cheap
transducers (2.5 UK pounds each) that although not exectly high
performance devices do, are quite capable of detecting resonably sized
objects up to about 2 meters away. These opperate at 40KHz.
The main feature of sound waves at these sorts of frequencies is that
they are reasonably directional and rapidly attenuated; which is why
they are so useful as obstacle detectors / range finders. I think you
need some quite significant power output (compared with the usual
transducers) to get any ranges above a few meters. As far as I
remember from my physics wave like propogation in air is quite
restricted in its frequency range, and the high the freq the more
rapid the attenuation.
Having said all that I'm still quite keen on using sound waves for all
sorts of things. We have a digitising device around that uses a sparc
to generate a sound pulse which is timed by two sensors about 50cm
apart to give position. These used be quite popular a few years back
as low cost digitising tablets; I guess mice etc killed them off.
I've also seen it suggest on a TV sience program (tommorows world),
that a similar principle could be used on a larger scale. They showed
a room with microphones in the corners. Using these a clicking
transmitter could be traced in 3D to an accuracy of about 1cm by
timing the arival of the pulses. Trouble is I can't remember who was
doing this, sounded like a system that would be fun to setup. I guess
in many environments the problems of interfering sound sources would
be quite significant though.
Anyone else got ideas on the use of sound. I get the feeling that like
other senses sound has been a little neglected in the rush to show
what large amounts of computing power vision requires.
--
-- Michael Stevens --
Dept. of Engineering Univerity of Warwick, Coventry, England