Re: Dropping marker signals?

[Tim Vanderhoek <vanderh@ecf.utoronto.ca>]

On Wed, Dec 29, 1999 at 01:08:21AM -0500, Joyce Poon wrote:
> 
> Anywho, I am slightly concerned about the structure of this extra
> transmitter.  Firstly, its sensors have to be kind of "tall" (at the same
> height as the beacon we place on the enemy), unless we angle down the
> emitters on the beacon (but that is rather silly as it will unnecessarily
> limit our sensor range).

Indeed, that may be a disadvantage.

Two possible solutions: 1) only drop a retransmitter beacon when we
                           run into an obstacle.  When we run into
                           an obstacle, extend an arm of some sort
                           and try to attach the beacon onto the
                           obstacle (obviously the retransmitter
                           beacons are equipped with some type of
                           sticky material) at an appropriate
                           height.  I don't particularly like this
                           solution.  Too complex.
                        2) Make them as tall as we can within the
                           constraints of whatever ejection mechanism
                           we device and hope it's not a problem.
                           Remember that IR light does tend to
                           spread-out ... it's not a laser, after
                           all.  It's only necessary for the beacon
                           to sense the enemy when the enemy is
                           reasonably far from the beacon anyways,
                           so this distance will give the IR light
                           time to spread-out.  Obviously we'd want
                           to test our hypothesis that "it won't
                           be a problem" before investing too much
                           work into retransmitter beacons.



> Detector diodes cost as little as $.50 to $.70 each (ie may require some
> extra components), while modules (ie may require less extra components) 
> cost about $4 each. 

Okay, but they're cheap.  That's probably USD, but they're still
cheap.  We can throw a couple of emmitter/detector pairs around and it
won't suddenly blow the budget...  :-)

[USD == US dollars]


> > Evaluate based on the above model: Visualize three dots placed randomly
> > in a square.  Draw circles of some radius around each dot.  Choose two
> > dots and move them according to the model's rules for robot movement.
[...] 
> 	Wait a sec, the circles are having different meanings.  The circle

Naw, they were meant to have the same meanings.  I was perhaps trying
to oversimplify it.


> 	To unify everything -- may I suggest that all 3 circles represent
> reception ranges.  In fact, to simplify the model, remove the circle
> around E (ie let us not concern ourselves with the enemy's detection

Okay.


> > Now.....I admit that when I visualize this test using the above model,
> > the rebroadcast idea isn't quite as useful as my original expectations,
[...]
>                                                       However, with my
> modification to your thought experiment, we can see that the increase in
> probability is not that high unless we have a large sensor range to start

Where "large" is defined relative to the total size of the playing
field.

> actually being in a circle).  I don't know what circle sizes you were
> thinking of, but suppose we were thinking of the same circles, you

I intentionally avoided stating what size circle.  :)  It does make a
difference, though, as you note.

I'm rereading Malone's old email and he only states that "sensor range
has been one of the greatest variables in the past", without giving
any clue as to WHAT those ranges where.  Reading from the net, I've
seen ranges from millimeters to many many feet.  I don't have a clue
as to what size these circles should be!! . . . . . .


[On the topic of using dropped beacons as "markers", a-la Hansel and Gretel]
>
> 	Hmm... in fact, why can't the computer just mark down where we
> made sharp turns or where we encountered obstacles on the map.  I think
> that would be easiest.

The computer could just mark down every turn and every obstacle,
except that my concern is that the feedback from the speed of the
wheels will not always be completely accurate.  Perhaps this is an
unjustified concern.  This concern, however, was the motivation for
the "dropped beacons" idea and the idea of using these beacons as
retransmitters was merely a secondary "hey, that's cool" idea.

I'd be much happier if I were convinced that the speed feedback will
always be highly accurate.  Certainly it's possible to have highly
accurate feedback -- the SR04 robot appears to have been extremely
accurate, but that doesn't mean it will work for us.  Our robot would
also be working under less favourable conditions wrt to speed feedback
than the SR04 robot worked under.  We could try to get Malone to give
us some idea of the accuracy we can expect.  I'd like to believe that
it'd be very accurate, but my concern is for the case where it isn't
so accurate...

>                        Moreover, dropping a beacon behind an obstacle is
> rather silly (esp if we are using IR) since the obstacle will probably
> block the signal. 

Not necessarily.  I seem to recall that both the Genesys and SR04
robot-makers expressed surprise at the number of different materials
that IR would pass through.


> 	Ok, I will assume that extra beacons will only be used for enemy
> detection.  Again, how many of these extra beacons can we carry?  (This
> goes back to the question I raised about the physical structure of the
> beacon.)  Since these transmitters will be kind of "tall", I envision that
> it will be difficult to drop them on the fly without something bad
> happening to them (like falling over).

Have you ever seen those little children's toys that have a rounded
bottom end with a weigth?  When you set them down, they automatically
stand up.  The weight at the bottom causes them to roll over until the
weight has reached the lowest position it can reach, which just
happens to correspond with the top of the toy being at the highest
position it can reach.

I'm still thinking about a good way to eject the beacons.  This is
Cindy's area.  :-)  Propose a method, Cindy.  :)

>                                        For enemy detection, ideally, the
> beacon should be placed in an open area to maximize its sensory range
> (although one can raise the question of why our robot will be near the
> open space to receive the signal emitted from the beacon particularly when
> we are in evade mode). 
 
[Why would our robot be near open space when in evade mode]

	Because open space has the largest number of directions in which we
	can run?

In general, though, I see this as something being useful when in
attack mode.  While in evade mode, we don't want to be running all
over the place, looking for the enemy ("looking for trouble")!

While in attack mode, open space may be the best place to be, since
that has line-of-sight IR transmission paths to the largest number of
places.  Obviously we wouldn't be able to say in "open space" for all
eternity, though.


> 	Maybe the best mode of transmission for these extra beacons should
> be radio signals.  We can have different frequencies of radio waves so
> that each frequency corresponds to a certain beacon.  However, I did a
> cursory browse on the web today and I couldn't find anything on tracking
> down the origin of a radio signal.  However, if the comp can somehow mark
> what beacon we dropped at a certain location, we can track it down again. 

Not being able to track down the origin of a signal somewhat defeats
the "aid in map construction/reliability/error-detection" purpose.
However, as you note, it has some major advantages wrt to information
transmission: radio waves pass through just about anything.  If the
enemy comes near a retransmitter beacon, that retransmitter beacon
will be able to tell our robot what's happened no matter where our
robot is.  _Very_ advantageous.

Concerns: 1) A full-fledged radio modem is probably more complexity
             than we want...  What type of radio transmitters could we
             use?
          2) Interference?  Probably not a big problem.
          3) Hm...  That's about all I can think of at the moment.  :)


> 	Furthermore, if our beacon (the one on E) also emits radio waves
> and the rebroadcaster detects radio waves of a certain intensity (ie
> within a certain range), then there is no need to make the rebroadcasters
> "tall."

[Put the period outside the quotes unless you're quoting a whole
 phrase of which the period is part.  This has become accepted as
 "new-style" or "logical-style" qouting by the OED.]

If our beacon (the one on E) emits radio waves, there probably is no
need for rebroadcasters at all!  My understanding is that radio waves,
when given sufficient power, can easily travers the whole distance
across the playing field, and through most obstacles.

>         As a side note, such a system will not be ideal for our robot
> itself since I have not yet found a feasible way of tracking down sources
> of radio waves. 

Hm.  Determing the direction from which a radio signal is coming is
fairly easy if you can shield radio signals!  Simply create a
directional reciever by shielding it from reception of radio waves
from all but one direction.  Just like a styrofoam cup with an
antennae inside the cup, for example.

Except that styrofoam does not shield radio signals, afaik.  :-)

How about microwave beacons?  It's pretty easy to build a microwave
sheild, for that matter.  :)

We were originally advised against using a radio beacon system by some
fourth-years because they were concerned that the circuitry involved
in building a radio transmission/reception system would be too
complex.  I don't know whether this is true or not.

Does your catalague list anything appropriate?


> 	Yeah... I thought of that before (I was thinking of all the
> different makes of VCRs and TVs; their signals don't get mixed up)... but
> the catalog says nothing about IR communications...  I will read the stuff
> on that Rice University website tomorrow. 

The billhir.html from Seattle Robotics discusses pulsed IR at a
frequency of 833Hz.


> > Oi.  Why not make the beacons into little helicopters while we adding
> > the rotating glass/plexiglass... 
> 
> 	Actually, while I was in the shower today, I was thinking why not
> make these beacons move.  Instead of dropping them, we could deploy them. 

Do you think that's legal?  (I guess why not?  ;-)

It's an interesting idea, and there certainly must be some way to
utilize it without too great complexity.  After all, the millirobots
people are expected to make dozens of their little bots, so it
couldn't be too hard for us to make a couple, especially if we don't
have the same constraints that they do.


> <please excuse the screwy brackets in the next paragraph... I don't know
> what went wrong here... the following paragraph is taken from Tim's last
> email>

I fixed them for you.

>>   For now, suffice it to say that having known stationary
>> beacons/markers would help to prevent the map from drifting as small
>> sensor errors accumulate (imagine trying to follow a map whose
[...] 
> 	As long as things are close enough the actual situation, all will
> be well. 

And therein lies my primary concern!  :)

Although, if we can work-out a way to use them as effective
rebroadcasters, that'd be pretty cool, too.

Because of the speeds our bot will (hopefully) travel at, a sudden
full change of direction, for example, could wreak havoc on the speed
feedback system.  Depending on the size of the motors we choose,
sufficiently large motors could cause the wheels to slip on the floor.
Quick changes in direction are tricky to determine --- neither Hall
sensors nor optic couples provide a "negative count" when the wheel
turns backwards.  (Although for most cases, we can simply assume the
wheel is turning in the direction we told it to, provided we slowed
the wheel to a stop before telling it to switch directions....but that
may not for all cases be entirely a straightforward thing).

Once the robot is built in main, it's easy to write a program that
would test the actual ability of the speed-feedback system to work
accurately.


> > If we're limited to one IR frequency (ugh!), then in the worst case
> > multiple lines of communication could be multiplexed over the single
> > channel using pulsed IR broadcasts at a different prime frequency for
> > each secondary communication line (tricky to get quite right, I
[...]
> 	I think I understand... you are saying that we can have different
> frequencies of IR pulses for each extra communication line we need. 

In fact, it's not even an original idea.  The billhir document from
Seattle Robotics suggests that secondary modulation is essential for
a Sharp IR detector to work effectively at all.

Of course, this is one of those ideas that sounds easy to write, but
in practice, interference and such would require a fairly smart
algorithm to filter-out the noise.  Needless to say, using different
IR wavelengths for different sensors is the preferred solution.


> Direct signal amplification may confuse our robot (since it will be
> possible for the robot to sense 2 IR signals in 2 directions at the same
> time). 

Not true.  If the robot is receiving two (or more) IR signals from
two (or more) directions, it can cross-check these directions with
its memory of where it put retransmitter beacons.  Presumably one
of the two IR signals is coming from a retransmitter beacon so the
computer can work out which signal should be ignored and which
signal must be the real enemy.

It's possible to work-out more complex scenarios where the robot still
becomes confused.  These can be avoided by not laying two
retransmitter beacons too close to each other.  It may not always be
easy to define "too close", however.


> > I don't think there's any reason it should complicate the design
> > significantly at all. 
> 
> 	If you say so... 

:-)


> > Some amount of testing to determine the correct heights and angles for
> > sensors and transmitters, (remember too that the transmitter will
> > potentially be placed at a varying height depending on the enemy robot's
> > constructors), but that would seem to be about all. 
> 
> 	There is a 5 cm height range for the beacon placement.  We could
> stick some mirrors/reflective material around the IR emitter to make the
> signal "wider." 

That might be a really good idea.  :-)  Preferrably some sort of
material that would diffuse the signal appropriately.


-- 
Signature withheld by request of author.