User Pass
Home Sign Up Contact Log In
Forum > Discuss GLB Issues With Catch22 > NOT A BUG - CLOSED > Breakaways ... why can't these players figure out the best way to run toward the endzone?
Page:
 
Link
 
I guess that leaves the question of Linebackers. Is there a way we could layer vision for them? Something that would identify the biggest threat and something that would identify a clear route? I mean, I don't have a good idea obviously but maybe someone smarter does.
 
Jose Bagg
offline
Link
 
Originally posted by Bort
I should try to find a way to post test games for scrutiny. Of course, I'm using real players/teams so that might piss some people off.


You can use any player on my team or my whole team for that matter.
 
Randy03Texags
offline
Link
 
Yeah when I built my autonomous robot it had two loops for the pathfinding the primary loop was to always travel in the direction of the acustic beacon, which it located by using differentiators in a microphone array, but the object avoidance subroutine (FPGA so really it was all parallel) used ultrasound sensors to avoid objects .. to the robot always moved towards the goal unless there was an object to avoid .. very similar to your two loop RB avoidance.
 
Pensfan
offline
Link
 
Originally posted by Randy03Texags
Yeah when I built my autonomous robot it had two loops for the pathfinding the primary loop was to always travel in the direction of the acustic beacon, which it located by using differentiators in a microphone array, but the object avoidance subroutine (FPGA so really it was all parallel) used ultrasound sensors to avoid objects .. to the robot always moved towards the goal unless there was an object to avoid .. very similar to your two loop RB avoidance.


Yeah! Me too!


 
Jose Bagg
offline
Link
 
I did the same thing with my robot O-servo.
 
lcplcombat
offline
Link
 
Originally posted by Bort
Good news is that somebody made an actual good suggestion on a fix for running pathfinding. I'm working with it right now in testing.

It's basically a double pass on the "best path" algorithm: one with a long vision distance for a general "avoidance vector" and one with a short, 3-5 yard distance for a "juke vector." Add 'em together, with the juke vector at a stronger strength than the avoidance vector, and it seems to get a pretty good path that really takes the runner's vision into account quite well.


Give the person who suggested the new idea some Flexpoints!!

Last edited Aug 19, 2008 09:46:07
 
Link
 
Originally posted by Bort
I should try to find a way to post test games for scrutiny. Of course, I'm using real players/teams so that might piss some people off.


I have 2 relatively well built CBs lvls 26 and 29 that you are more than welcome to use however much you need them and in any capacity.
Id like to see these coverage kinks worked out asap and if I can help speed the process I would do whatever I can.
 
drifter
offline
Link
 
Originally posted by Bort
Good news is that somebody made an actual good suggestion on a fix for running pathfinding. I'm working with it right now in testing.

It's basically a double pass on the "best path" algorithm: one with a long vision distance for a general "avoidance vector" and one with a short, 3-5 yard distance for a "juke vector." Add 'em together, with the juke vector at a stronger strength than the avoidance vector, and it seems to get a pretty good path that really takes the runner's vision into account quite well.


I used to get mentioned by name, now I'm just somebody?

Harrumph
 
drifter
offline
Link
 
Originally posted by drifter
Originally posted by Bort

Good news is that somebody made an actual good suggestion on a fix for running pathfinding. I'm working with it right now in testing.

It's basically a double pass on the "best path" algorithm: one with a long vision distance for a general "avoidance vector" and one with a short, 3-5 yard distance for a "juke vector." Add 'em together, with the juke vector at a stronger strength than the avoidance vector, and it seems to get a pretty good path that really takes the runner's vision into account quite well.


I used to get mentioned by name, now I'm just somebody?

Harrumph




For posterity

Originally posted by drifter
Bort - Tough to speak to this without knowing exactly how the parsing works, but it seems like there should be two circles of awareness. One for pathfinding and one for "move making".

I'm guessing that the current problem is that the player makes a pathfinding decision based on what it perceives as the best option, yet as players move around the field, another option may suddenly be better and then a second later, yet another option is seen as better.

If you have two circles of awareness, set the pathfinding circle to make a determination, say every 5 yards with current direction being weighted more favorably as well as a decreasing weighting ranging form 0 degrees to +/- 110 degrees (I'm guessing there's something like the second part of this already in place). This keeps the player from stutter stepping to change directions to constantly changing options when players are some distance from them.

The second circle of awareness, is set up pretty much as it is now with a very small radius, allowing the player to make "move" decisions more quickly based on players within a close range and occurs very quickly (at whatever rate it's set to make that determination now).

Both circles are operating concurrently, with one making a decision either every 5 yards or every so many ms and the other closer range one making constant determinations.

Would that work?


 
usafdirtboyz
offline
Link
 
The dots are not exact!
 
222
offline
Link
 
Originally posted by drifter
Originally posted by drifter

Originally posted by Bort


Good news is that somebody made an actual good suggestion on a fix for running pathfinding. I'm working with it right now in testing.

It's basically a double pass on the "best path" algorithm: one with a long vision distance for a general "avoidance vector" and one with a short, 3-5 yard distance for a "juke vector." Add 'em together, with the juke vector at a stronger strength than the avoidance vector, and it seems to get a pretty good path that really takes the runner's vision into account quite well.


I used to get mentioned by name, now I'm just somebody?

Harrumph




For posterity

Originally posted by drifter

Bort - Tough to speak to this without knowing exactly how the parsing works, but it seems like there should be two circles of awareness. One for pathfinding and one for "move making".

I'm guessing that the current problem is that the player makes a pathfinding decision based on what it perceives as the best option, yet as players move around the field, another option may suddenly be better and then a second later, yet another option is seen as better.

If you have two circles of awareness, set the pathfinding circle to make a determination, say every 5 yards with current direction being weighted more favorably as well as a decreasing weighting ranging form 0 degrees to +/- 110 degrees (I'm guessing there's something like the second part of this already in place). This keeps the player from stutter stepping to change directions to constantly changing options when players are some distance from them.

The second circle of awareness, is set up pretty much as it is now with a very small radius, allowing the player to make "move" decisions more quickly based on players within a close range and occurs very quickly (at whatever rate it's set to make that determination now).

Both circles are operating concurrently, with one making a decision either every 5 yards or every so many ms and the other closer range one making constant determinations.

Would that work?




You sir deserve a cookie!
 
drifter
offline
Link
 
Originally posted by 222
Originally posted by drifter

Originally posted by drifter


Originally posted by Bort



Good news is that somebody made an actual good suggestion on a fix for running pathfinding. I'm working with it right now in testing.

It's basically a double pass on the "best path" algorithm: one with a long vision distance for a general "avoidance vector" and one with a short, 3-5 yard distance for a "juke vector." Add 'em together, with the juke vector at a stronger strength than the avoidance vector, and it seems to get a pretty good path that really takes the runner's vision into account quite well.


I used to get mentioned by name, now I'm just somebody?

Harrumph




For posterity

Originally posted by drifter


Bort - Tough to speak to this without knowing exactly how the parsing works, but it seems like there should be two circles of awareness. One for pathfinding and one for "move making".

I'm guessing that the current problem is that the player makes a pathfinding decision based on what it perceives as the best option, yet as players move around the field, another option may suddenly be better and then a second later, yet another option is seen as better.

If you have two circles of awareness, set the pathfinding circle to make a determination, say every 5 yards with current direction being weighted more favorably as well as a decreasing weighting ranging form 0 degrees to +/- 110 degrees (I'm guessing there's something like the second part of this already in place). This keeps the player from stutter stepping to change directions to constantly changing options when players are some distance from them.

The second circle of awareness, is set up pretty much as it is now with a very small radius, allowing the player to make "move" decisions more quickly based on players within a close range and occurs very quickly (at whatever rate it's set to make that determination now).

Both circles are operating concurrently, with one making a decision either every 5 yards or every so many ms and the other closer range one making constant determinations.

Would that work?




You sir deserve a cookie!


I do enjoy cookies
 
islander1
offline
Link
 
Originally posted by Cruzi
I saw someone joke about high vision creating problems. Well, I thought it was a joke. This is a bad bug.

Is there a way to add in a player's field of vision to make it where he can't see out the back of his helmet? Might help.


...unless he's watching himself on the Jumbotron, sees someone gaining, and just panics. Must be a confidence issue. :-/


My linebacker has 50 vision, and he was more effective with half that vision. Other linebackers with less vision on my own team make more decisive plays.
 
Bort
Admin
offline
Link
 
Originally posted by drifter
Originally posted by Bort

Good news is that somebody made an actual good suggestion on a fix for running pathfinding. I'm working with it right now in testing.

It's basically a double pass on the "best path" algorithm: one with a long vision distance for a general "avoidance vector" and one with a short, 3-5 yard distance for a "juke vector." Add 'em together, with the juke vector at a stronger strength than the avoidance vector, and it seems to get a pretty good path that really takes the runner's vision into account quite well.


I used to get mentioned by name, now I'm just somebody?

Harrumph


I knew it was somebody smart; just couldn't remember who!

Cookie for you.
 
that1guy
offline
Link
 
Originally posted by Bort
Good news is that somebody made an actual good suggestion on a fix for running pathfinding. I'm working with it right now in testing.

It's basically a double pass on the "best path" algorithm: one with a long vision distance for a general "avoidance vector" and one with a short, 3-5 yard distance for a "juke vector." Add 'em together, with the juke vector at a stronger strength than the avoidance vector, and it seems to get a pretty good path that really takes the runner's vision into account quite well.


is there a way that this "best path" algorithm can be used to help defenders take better angles of attack?

http://goallineblitz.com/game/replay.pl?game_id=125654&pbp_id=4805282

not the first time my FS has taken a poor angle and with 50 vision and 5 superior vision the guy is far from blind. there have been other plays where the FS is in persuit and seems to run to directly parrallel to the guy he's chasing, almost as if he can only opperate on straight lines.
http://goallineblitz.com/game/replay.pl?game_id=84788&pbp_id=2661714
Last edited Aug 19, 2008 18:38:51
 
Page:
 


You are not logged in. Please log in if you want to post a reply.