🎉 Celebrating 25 Years of GameDev.net! 🎉

Not many can claim 25 years on the Internet! Join us in celebrating this milestone. Learn more about our history, and thank you for being a part of our community!

Update

Started by
26 comments, last by khawk 20 years, 11 months ago
quote: Original post by Anonymous Poster
is there a way to get your current position?


if i understand how the bot interface is going to work correctly, i don''t believe that there will be a function that returns your definite world-oriented coordinates. any information your bot receives about coordinates of perceived objects in the arena will be relative to the bot itself (just like in real life! when you see a tree you don''t know it''s absolute, GPS coordinates, you only know it''s relative location to yourself and other objects you can see around you). in my opinion this is what is going to make this contest so challenging! traditional game ai techniques wont be so easy to apply.

however, i could be totally off base here...this is just how i think the interface is going to work.


A headache, ancillary; an hourglass auxiliary.

"something witty, blah, blah, blah"
--That One Witty Guy (remember? that one dude?)


------------------------------------------------------- A headache, ancillary; an hourglass auxiliary."something witty, blah, blah, blah" --That One Witty Guy (remember? that one dude?)(author of DustBot)
Advertisement
Dittos to Xtremehobo. Why are you on probation?


(Stolen from Programmer One)
UNIX is an operating system, OS/2 is half an operating system, Windows is a shell, and DOS is a boot partition virus
Yeah do you get absolute positions/direction/velocity or any way of finding them relative to your start situation. Could have drastic implications on by map representation.

ps what IS probation & why are so many people on it suddenly including at least 1 moderator and a member of staff?!

pps & this is a really wierd forum where all but one post are from GDNet staff members!

[edited by - d000hg on July 28, 2003 7:57:45 AM]
quote: Original post by d000hg
Yeah do you get absolute positions/direction/velocity or any way of finding them relative to your start situation. Could have drastic implications on by map representation.


well, it should be relatively straightforward to find positions/direction/velocity relative to your start point. you just arbitrarily say that your start point is 0,0 and keep track of the bots displacement from this point from there on out. then, to find any observed object''s coordinates in this coordinate system you just add the bot''s coordinates to the relative coordinates of the observed object.

this should work, right? does anyone see any problems with this?

perhaps a better question: is this how the interface is really going to work (khawk?)? i mean, are we not going to have access to the bot''s absolute world-coordinates?

well i guess we should just wait and see for ourselves...but im just an impatient american monkey!





A headache, ancillary; an hourglass auxiliary.

"something witty, blah, blah, blah"
--That One Witty Guy (remember? that one dude?)


------------------------------------------------------- A headache, ancillary; an hourglass auxiliary."something witty, blah, blah, blah" --That One Witty Guy (remember? that one dude?)(author of DustBot)
quote: Original post by drreagan
you just arbitrarily say that your start point is 0,0 and keep track of the bots displacement from this point from there on out. then, to find any observed object''s coordinates in this coordinate system you just add the bot''s coordinates to the relative coordinates of the observed object.

this should work, right? does anyone see any problems with this?

It should work, as long as KHawk isn''t deliberately putting in distance and angle measurement errors (ie, in real life, you can''t tell exactly how far you''ve moved or turned without special equipment like a GPS system). If measuring errors are modelled, then it''ll be a bit harder. If you want to make an internal map then you''d have to test object positions to see whether they''re close to an object you already know about, and ignore the new position (or if you want, change the stored position to be the average of the two).

It''s quite a common problem in real-world robotics, because of measurement errors caused by wheels/tracks slipping.

So now the question simply becomes, does the system model measurement error? And if it isn''t at the moment, should it be added for later contests? (I''m not sure whether I''d like it to be added later, but imho it should be left out for this first contest)

John B
The best thing about the internet is the way people with no experience or qualifications can pretend to be completely superior to other people who have no experience or qualifications.
quote: Original post by d000hg
ps what IS probation

A warning not to do anything stupid. Like a note to say: We''re watching you.

John B
The best thing about the internet is the way people with no experience or qualifications can pretend to be completely superior to other people who have no experience or qualifications.
The only thing I am wondering about is how the bot can ''see'' walls.

And if there was a random error, I doubt it would be large enough to cause it to look like two objects are in the same place.
quote: Original post by Xgkkp
The only thing I am wondering about is how the bot can ''see'' walls.

do you mean in terms of data, or something else?

i was thinking he would place the walls in the data coming from the GetObjectInSight method. maybe enumerated types, maybe RTTI, maybe leaving it up to the bot to determine what type of obstacle it was, assuming that was even needed or wanted. i''m a real noob at all of this - is there something special we need to know about walls as opposed to trees?
quote: Original post by flipper76108
quote: Original post by Xgkkp
The only thing I am wondering about is how the bot can ''see'' walls.

do you mean in terms of data, or something else?

i was thinking he would place the walls in the data coming from the GetObjectInSight method. maybe enumerated types, maybe RTTI, maybe leaving it up to the bot to determine what type of obstacle it was, assuming that was even needed or wanted. i''m a real noob at all of this - is there something special we need to know about walls as opposed to trees?


Trees could be represented as a ''point'' or ''radius'' or even an angle from the bot''s direction, distance and width from the bots point of view. Walls are more complicated because they have a start and an end, the bot can see ''slices'' at different angles and can see more than one of them. Things like trees presumably ''look'' the same no matter what angle you are at. It''ll be interesting to see how He''s done it.
quote: Original post by Xgkkp
And if there was a random error, I doubt it would be large enough to cause it to look like two objects are in the same place.

The problem is not two objects looking like they''re in the same place. The problem is after you''ve moved around a lot, with measurement errors, an object that you saw when you started out could appear to have moved, and so you might end up with two copies of it in your internal map.
In other words, it''s not two objects looking like they''re in the same place, it''s one object looking like it''s in two places.

John B
The best thing about the internet is the way people with no experience or qualifications can pretend to be completely superior to other people who have no experience or qualifications.

This topic is closed to new replies.

Advertisement