AIPlayer::getAiMove()
by John Klima · in Torque Game Engine · 06/19/2005 (11:45 am) · 2 replies
Hi all,
i'm playing around with aiPlayer, trying to get a number of bots to move around. first off, was wondering why onMoveStuck was never getting called.
in the code it checks to see if it is stuck
problem is, no where in code is mLastLocation ever being set. why not?
so i changed it to this:
which fires the callback, sometimes. i expected that at any moment from frame to frame a bot is in the same position, the callback should be thrown. but i can look at the bots at runtime, they sit there doing nuthin, but the callback is not thrown. why????
thanks,
j
i'm playing around with aiPlayer, trying to get a number of bots to move around. first off, was wondering why onMoveStuck was never getting called.
in the code it checks to see if it is stuck
if (location == mLastLocation) { ...problem is, no where in code is mLastLocation ever being set. why not?
so i changed it to this:
// We should check to see if we are stuck...
if (location == mLastLocation) {
throwCallback("onMoveStuck");
//not sure what state it should be, but if it's stuck
//i don't think i want it to be ModeStop, what was here originally.
//regardless, behavior is unclear
mMoveState = ModeMove;
} else {
mLastLocation = location;
}which fires the callback, sometimes. i expected that at any moment from frame to frame a bot is in the same position, the callback should be thrown. but i can look at the bots at runtime, they sit there doing nuthin, but the callback is not thrown. why????
thanks,
j
#2
the initial Q still remains though, why does the virgin AIplayer code come "broken." without the end-user actually setting mLastLocation in the c++ code, you will never get the stuck callback to fire. i pity the torque user who cant modify and re-compile the source.
here's what i got in my AIplayer now:
06/20/2005 (3:34 pm)
I worked it out by juggling things around in the aiMove method, and by doing some diff math rather than simply mLastLocation = location. this way the stuck callback occurs at reasonable frequency.the initial Q still remains though, why does the virgin AIplayer code come "broken." without the end-user actually setting mLastLocation in the c++ code, you will never get the stuck callback to fire. i pity the torque user who cant modify and re-compile the source.
here's what i got in my AIplayer now:
...
// Move towards the destination
if (mMoveState == ModeMove || mMoveState == ModeStuck) {
F32 xDiffLast = 0;
F32 yDiffLast = 0;
//first check if we are still in stuck mode from a previous cycle
if (mMoveState == ModeStuck){
xDiffLast = mFabs(mStartAvoidLocation.x - location.x);
yDiffLast = mFabs(mStartAvoidLocation.y - location.y);
if (xDiffLast + yDiffLast > mMoveTolerance *10){
//okay, i'm far enuf away from my stuck point to call unstuck
//which will re-try moving towards the original goal point
throwCallback("onMoveUnStuck");
mMoveState = ModeMove;
}
}
//now check if we are newly stuck
xDiffLast = mFabs(mLastLocation.x - location.x);
yDiffLast = mFabs(mLastLocation.y - location.y);
// We should check to see if we are stuck...
//fiddle with "20" to adjust sensitivity
if ((xDiffLast + yDiffLast) *20 < mMoveTolerance ){
//hold onto my initial stuck location
mStartAvoidLocation = location;
//script picks a temporary goal point to avoid
throwCallback("onMoveStuck");
mMoveState = ModeStuck;
}
mLastLocation = location;
}
... 
Torque Owner Mark Storer