Robot ``Easter Egg Hunt'' discussion digest

Next discussion page
Date: 13 May 1999
DJ: Darrell Johnson
BH: Ben Hitchcock
KH: Ken Huntington
SB: Steven Bolt
Back to the Egg Hunt home page

Changing the Contest rules and Playground definitions?

KH> I decided that I would make communications for this endevour
KH> separate from all others as anything pertaining to the hunt 
KH> may be freely quoted from :)
KH> As indicated above these are *thoughts* and this is *your*
KH> Playground and I am comfortable with that :)

SB> I very much appreciate your thoughts, I'm quite sure I didn't
SB> get it (entirely :) right to begin with, and I'll be happy to
SB> improve the Egg Hunt!

SB> But I think it would not be correct to view this as my
SB> Playground to do with as I please. The best way to deal with
SB> changes, I think, is to give all Egg Hunters the right to
SB> propose and to veto them. Of course that rule is effective
SB> only if we know who is (and who isn't) an Egg Hunter. This
SB> digest is probably the best criterion. Once someone has
SB> contributed to it, making him/herself known in public as one
SB> who is seriously interested in developing an Egg Hunting robot,
SB> that person is an Egg Hunter with the rights as described
SB> above. Objections to any proposed change have to be filed
SB> within let's say three weeks.

The contest

SB> That brings me to the contest itself. I would prefer not to
SB> organize it, at least not in the usual way, because I want to
SB> participate! But I wonder if we can't make the contest itself
SB> as symmetrical and democratic as the Playground definition.
SB> At a very early stage I was mailing about Egg Hunting with only
SB> one other person, Ben Hitchcock. This was said about the contest:

> BH> Put me down as competing. Even if there are no prizes, and
> BH> only one other bot enters, I want to have a go at this one.
> BH> I'll gladly post my entry off to wherever the contest is, with
> BH> the return postage inside.

> SB> You'll be competing against me, at least :)
> SB> And should that still be the situation when our Egg Hunters are
> SB> contest-ready, then I suggest we mail our `bots to each other.
> SB> With some luck the weight will be equal, which would make it
> SB> unnecessary to include return postage.

SB> Two participants would both pay for shipment of two robots, to
SB> have a contest round at each location and have their own `bots
SB> returned. Quite doable, as each participant has his/her own
SB> Playground anyway.
SB> With for instance five participants, we could still have that
SB> symmetry. Each of us would have to pay five shipment fees, but
SB> that is perhaps not too high a price for the fun of running your
SB> own contest round. Results would be averaged to decide the final
SB> winner.
SB> If the number of participants grows beyond five or so, we can't
SB> have all `bots travel to all locations. Instead we could split
SB> them up in groups, to have group winners and a final.

SB> This can work only if - when the robots are ready for the contest
SB> - all participants trust each other to handle and return the
SB> robots with care and speed. I rather hope that will be the case.
SB> But if by that time we should prefer a centrally organized
SB> `mail-in' contest, then if all else fails I'll organise it, even
SB> if that means my Egg Hunter can't compete. I will even find some
SB> interesting prizes :)
SB> Am I talking some sort of sense here?

Obstacle min/max. diameter

KH> and second, there is no minimum or maximum diameter given
KH> for the obstacles.

SB> No need: "Ten obstacles: cylindrical, 2cm high, matt finish,
SB> kept in position with a central pin. The distance between
SB> them and between any obstacle and the border is 8cm or more.
SB> They may be anywhere on the Playground, except in the
SB> gathering area." Imho that sufficiently defines the obstacle
SB> situation.

KH> Yes, that's what I read and if that's the way it is then,
KH> vision, infrared radar, or sound echo location functions
KH> cannot be used to locate an obstacle as by the descriptions
KH> of the Playground and the Obstacle the diameter can be
KH> almost zero (just a pin) to 50 -8 -8 = 34cm. So touchy feely
KH> is the only sensor guaranteed to locate an Obstacle. IMHO,
KH> that is a 'must understand' limitation.

SB> I don't entirely agree with that. In practice, vision and
SB> other advanced sensors always need a tactile backup.
SB> Incorporating effective longer range sensors will give you
SB> an advantage, as they will see most obstacles and certainly
SB> the wall. You merely have to take into account that they may
SB> sometimes fail, which imho is a good thing to do anyway.

SB> But, if you feel that the obstacle diameter should be bound
SB> to a minimum and a maximum, I have no real objection. If you
SB> quantify your proposal, I'll ask the other known Egg Hunters
SB> for comment - and whether they'd like an informal `mailing
SB> list' to discuss Egg Hunt progress.
KH> 8) The minimum diameter of an Obstacle should be 2cm.

SB> If there is to be a minimum, then let's have a maximum as well.
SB> 20cm?
KH> I was only thinking of visibility for whatever sensor is used.
KH> 20cm is just fine.

SB> Summarizing, subject to approval by the other Egg Hunters:

SB> Ten obstacles: cylindrical, 2cm high, matt finish, kept in
SB> position with a central pin. The distance between them and
SB> between any obstacle and the border is 8cm or more. They may be
SB> anywhere on the Playground, except in the gathering area.
SB> Diameter between 2cm and 20cm.

Height of the Playground border

KH> 1) The border should be 5cm high as to be clearly visible
KH>    above all obstacles.

SB> I deliberately made it as high as the obstacles, to avoid
SB> clever but too specific tactile sensors, differentiating
SB> Eggs, obstacles and wall by being 1cm, 1.75 cm and 2.5 cm
SB> above the floor.
KH> Identifying what's what is important and I hadn't thought of
KH> that type of tactile sensing (height discrimination) but
KH> part of my preliminary design does involve a specific
KH> tactile sensor which, with a little logic, should be able to
KH> do just that. (just trying to use the KISS principle)

SB> I want to stimulate KISS, but avoid a too specific and
SB> entirely mechanical approach, as was used by David Miller
SB> for his Scarecrow (Fourth in the 1992 AAAI Mobile Robotics
SB> Competition: "A half-dozen relays, a couple of reostats, and
SB> eight switches," well ahead of several `bots using advanced
SB> vision systems and/or serious off-board computing power.)

SB> So I remain against height differences other than the one
SB> between the eggs and everything else. Another Egg Hunter has
SB> proposed to give the obstacles a color, other than black or
SB> white. Now if he or someone else comes up with a color
SB> discrimator, able to deal with (temporary) shades and
SB> environmental light intensity changes, that would be great!
SB> But I'm inclined to await that color discriminator, before
SB> changing the playground definition. As it stands, the
SB> obstacles merely don't have to be white.

SB> Btw, I think it is possible to discriminate between the flat
SB> walls and the cylindrical obstacles. My own Egg Hunter will
SB> be able to attempt that.

Black walls for the Egg Hunt gathering corner

KH> 2) The border should be matt black in the Gathering Area.

SB> Hmmm... Here is a summary of a conversation with Darrell
SB> Johnson:

> SB> How much higher would you want them?

> DJ> I'm not really sure how high they should be.. The higher
> DJ> they are, the more controlled the environment is..  5 cm
> DJ> walls might be high enough though..

> SB> That gives you a pretty small angle of view on that black
> SB> strip, when looking from the long end of the Playground:
> SB> atan(5/100) is less than 3 degrees. As obstacles may well
> SB> hide the lower 2cm, your sensor may have to live with 1.7
> SB> degrees. And what happens if somebody standing near the
> SB> Playground wears dark clothes?
> DJ> yeah, that might be asking for a bit too much precision..
> DJ> big bright or dark colored objects outside the Playground
> DJ> would most definately throw off the sensors. I'll have to
> DJ> play with my design and see what I can achieve before I ask
> DJ> too much more.. maybe I'll have to scrap the idea and go for
> DJ> the IR tracker.. 

> SB> No harm in considering alternatives. We'll need a good bit of
> SB> lateral thinking, to find those Eggs :)

KH> This is true but that infers that it can't be done or done
KH> easily. I'd hate to think that a good idea could be nipped
KH> in the bud by such logical arguments but there again, if the
KH> suggester still thinks that the idea has merit, logical
KH> arguments would not dissuade :)

Modulated IR Beacon

KH> 3) The beacon should be provided by the Egg Hunter.
KH> 4) The beacon should contain it's own power regulation.

SB> The current definition allows:

SB>    Participants may provide their own IR beacons, as long as
SB>    the average current consumption does not exceed 10mA at
SB>    5V, and the emitters are at a height of 3cm. A stabilized
SB>    5V source will be available.

SB> The 5V standard and available source helps limit the weight
SB> of the package to be mailed, and 5V is easy to supply. Do you
SB> feel that the voltage and power limitations are too
SB> restrictive?
KH> No, only that if a problem arises, there could be no doubt that
KH> the fault cannot be the provided power.

SB> Good point. Let's make that supply 9V unstabilized, with the
SB> current limit remaining at 10mA, if no one objects.

KH> 5) The beacon should be placed in the Gathering Corner.
KH> Restricting a beacon to the corner nearest to the center of the
KH> arc and the transmission restricted to 3cm above the floor would
KH> tend to keep the playing field level. The only reason I suggest
KH> that it might be better in the corner as opposed to on or outside
KH> the corner is to make it easier to create the beacon and it would
KH> not be influenced or made more difficult to construct by the
KH> height of the border should it be decided that its height should
KH> change at future date.

KH> 6) The beacon should occupy an area of less than 25cm2.

SB> As shown on the illustration, the beacon is not on the
SB> Playground floor. Should we worry about area taken up outside
SB> it, or state more clearly that it is outside?
KH> Talk about being clear :) That area was suggested if the beacon
KH> were allowed to be placed within the border in the corner.

KH> 7) The beacon should be optional.

SB> Seems to go without saying, but I'll be happy to add that
SB> phrase :) 

KH> Clarity for something that seems obvious. At CDC, the
KH> specification a proposed product is considered to be the single
KH> most important part of a project even though it is rarely
KH> signed off until well into the program.  It's as much for the
KH> customer as the designers. One must never allow the customer or
KH> designer to find a loop-hole to crawl through as closing it
KH> costs money and/or creates hard feelings.

SB> I do agree! And adding that phrase not only provides clarity,
SB> it stimulates lateral thinking. I guess I disregarded the
SB> option *not* to use that beacon, because it's central to all
SB> simple navigation methods I can think of.

SB> I'd prefer to keep the height of the construction (excluding
SB> beacon) at 2cm, as that makes the Playground easy to build and
SB> store. So I'm inclined to summarize this as:

SB> Modulated IR beacon: 
SB> Optional. Participants who wish to use an IR beacon must provide
SB> their own. The average current consumption must not exceed 10mA
SB> at 9V, and the emitters must be at a height of 3cm, at the center
SB> of the gathering arc (as shown). If the beacon is to stand in the
SB> gathering area (instead of outside the Playground), it must occupy 
SB> an area of less than 25cm2.  An unstabilized 9V source will be
SB> available.

SB> The above is again subject to approval by the other Egg Hunters.

The Number of obstacles and Eggs

KH> 9) The number of Obstacles has no significance as long
KH>    as the placement rules can be met.
KH> 10) The number of Eggs depends totally on the production
KH>     of the Easter Bunny.

SB> At present I (merely) presume that the number of Eggs and
SB> obstacles has a strong influence on the time a `bot needs to
SB> gather all Eggs. As that time is an important contest
SB> element, I'd like to keep these numbers fixed for the time
SB> being. We can allways throw in your 9 and 10 after the first
SB> contest, if everyone agrees.
KH> Just observations but your right especially about the number
KH> eggs. There must always be sufficient room in the gathering
KH> area for all the eggs so keep a close eye on the bunny.

Discussing details of the Egg Hunting robots

SB> Btw, I'm not sure that I'll be able to not share any details,
SB> being naturally blabby about my `bots :)

KH> Others may prefer to unveil just the finished product; TA
KH> Daaaa and here he/she/it is... I'll ponder some before
KH> getting blabby :)

SB> I'll ponder the idea of a little `mailing list' for a while.
SB> It should leave room for a little secrecy, yet allow free
SB> discussion...
SB> Perhaps the Egg Hunt will remain just outside BEAM proper. No
SB> problem, though I'd like to see and wouldn't underestimate a
SB> non-uC contender.

KH> I'd hate to be shown up by some dumb blond 'bot :)

SB> Somehow I'd rather like that to happen. It would be really cool if
SB> there was some utterly simple answer to the Egg Hunt puzzle. On
SB> paper, it already inspired a delightfully simple solution to the
SB> tactile sensor problem, and I believe I also have just the right
SB> (and rather different) `jaws' for the task...
KH> Now see you've piqued my interest again :)

SB> I'm beginning to wonder whether Egg Hunters on the drawing
SB> board should or shouldn't be discussed. Wouldn't discussing
SB> them speed up development, and help avoid a waste of time on
SB> bad ideas?

KH> That would depend on the individual. We have enjoyed
KH> discussing many ideas as may the others. It would depend on
KH> how sensitive we are when presenting a critique.

SB> And on whether secrecy and open discussion can be profitably mixed.

KH> Secrecy is not just to win 'The Prize', but there's also the
KH> feeling that you've been able to accomplish something by
KH> yourself. Sharing design ideas or listening to the ideas of
KH> others before the accomplishment might diminish the elation of
KH> success. Offering help, when asked, is great. Comparing notes
KH> and discussing strengths and weaknesses of a design after the
KH> contest can be when much of the fun and learning takes place.
KH> The second round of competition should be where the real design
KH> advancements are evident, IMHO. Then, of course, there can be
KH> no absolutely correct answers since there's more that one human
KH> involved :)

SB> Free talk about design details may speed up development and
SB> avoid time wasted on dead ends. On the other hand, a good idea
SB> may stifle the birth of a better one. Since my Egg Hunter is at
SB> present just a paper design - apart from the motors and wheels
SB> - I don't know for sure which parts are dead ends which are
SB> good ideas :) 
SB> So I'll discuss it in detail, as soon as I have hardware and/or
SB> tested schematics to show. That leaves some time for other
SB> ideas to ripe and should make the publication a more interesting
SB> read as well.

Back to the Egg Hunt home page