Predicting Collisions..?

I'm trying to make a flashy shielder. I know there are anti-push scripts out there, but making a shielder sounds fun ^^


However, I am having trouble coming up with an algorithm to predict collisions on a llSensorRepeat()..

The only one I've come up with is as follows:

vector velocity;
integer i;
integer k;
integer y;
float tesme;
integer test;
vector testpos;
vector testpos2;
vector newpos;
//
default
{
    state_entry()
    {
        llListen(0,"",llGetOwner(),"");  //If I say "shield" it rezzes the shield object
        llSensorRepeat("","",AGENT | ACTIVE,90,2*PI,.1);
    }
    on_rez(integer bla)
    {
        llListen(0,"",llGetOwner(),"");
        llSensorRepeat("","",AGENT | ACTIVE,90,2*PI,.1);
    }
    sensor(integer t)
    {
        for(i = 0; i < t; i++)
        {
            velocity = llDetectedVel(i);
            tesme = llVecMag(llGetPos());
            if(llVecMag(velocity) > 4)
            {
                //llSay(0,"Ahh!!"); - If it's not rezzing shields, i use this to see if its even working..
                testpos = llDetectedPos(i);
                testpos2 = llGetPos();
                newpos.x = testpos2.x - testpos.x;  //Find the distance between the shielder and the moving object
                newpos.y = testpos2.y - testpos.y;
                newpos.z = testpos2.z - testpos.z;
                if(velocity.x / newpos.x - llRound(velocity.x / newpos.x) < .3)
                {
                    if(velocity.y / newpos.y - llRound(velocity.y / newpos.y) < .3)
                    {
                        if(velocity.z / newpos.z - llRound(velocity.z / newpos.z) < .3) //If all the distances in x, y, z are divisable by the Velocity, then it must be 'on line'... (supposadely)
                        {
                            //llPushObject(llDetectedKey(i),-velocity * 5, ZERO_VECTOR, FALSE);
                            if(llVecMag(llDetectedPos(i)) > llVecMag(llGetPos())) { //a crappy way of seeing if its going the right direction
                newpos = llGetPos() + ((llDetectedPos(i) - llGetPos()) / 5);
}
                            else
                            {
                                newpos = llDetectedPos(i) + ((llGetPos() - llDetectedPos(i)) / 5);
                            }
                            
                            
                            llRezObject("Shield", newpos, ZERO_VECTOR, ZERO_ROTATION, 42);
}
}
}
            }
        }
    }
                    
    touch_start(integer total_number)
    {
        //
        llSay(0, "Touched.");
    }
    listen(integer chan, string name, key id, string m)
    {
        if(m == "shield") {
            llRezObject("Shield", llGetPos() + <0,2,0>, ZERO_VECTOR, ZERO_ROTATION, 42);
        }
    }
}

I'm no math genius.. This method works, kind of.. but its way too slow. The object always hits the shielder before it generates a shield.

Any suggestions well appreciated ^^

Doing my unanswered post forum rounds for the day. Here's my reaction.

Your bottleneck appears to be the sensor event code. Instead of using all of that, try something like this.

sensor(integer t)
    {
        for(i = 0; i < t; i++)
        {
            velocity = llDetectedVel(i);
            if(llVecMag(velocity) > 4)
                    llRezObject("Shield", llGetPos() + (3 * llVecNorm(llDetectedPos(i) - llGetPos())), ZERO_VECTOR, ZERO_ROTATION, 42);
        }
    }

Much simpler.

the problem with your suggestion Jeffery is that vecmag doesnt determine wether or not an object is coming toward you or going away from you. It triggers shield rez wether or not an object is coming or going. I have yet to work this kink out.

True, but the problem here was speed, not necessarily accuracy. Lower the sensor range and it should be of no consequence.

Of course, if you want to really nail it, you can always "fake" it:

llVecNorm(llDetectedPos(i) - llGetPos()) - llVecNorm(llDetectedVel(i))

Then run the whole vector or select components against a certain threshhold value (say, less than 0.5) - but again, you lose a little speed. Still faster than the above, though.

thank you for this

you know, its still slower than the sion shield. fancy that!