In my 2D game, I have stationary AI turrets firing constant speed bullets at moving targets.

So far I have used a quadratic solver technique to calculate where the turret should aim in advance of the target, which works well (see Algorithm to shoot at a target in a 3d game, Predicting enemy position in order to have an object lead its target).

But it occurs to me that an iterative technique might be more realistic (e.g. it should fire even when there is no exact solution), efficient and tunable - for example one could change the number of iterations to improve accuracy.

I thought I could calculate the current range and thus an initial (inaccurate) bullet flight time to target, then work out where the target would actually be by that time, then recalculate a more accurate range, then recalculate flight time, etc etc.

I think I am missing something obvious to do with the time term, but my aimpoint calculation does not currently converge after the significant initial correction in the first iteration:

import math

def aimpoint(iters, target_x, target_y, target_vel_x, target_vel_y, bullet_speed):

    aimpoint_x = target_x
    aimpoint_y = target_y
    range = math.sqrt(aimpoint_x**2 + aimpoint_y**2)
    time_to_target = range / bullet_speed
    time_delta = time_to_target
    n = 0

    while n <= iters:
        print "iteration:", n, "target:", "(", aimpoint_x, aimpoint_y, ")", "time_delta:", time_delta
        aimpoint_x += target_vel_x * time_delta
        aimpoint_y += target_vel_y * time_delta
        range = math.sqrt(aimpoint_x**2 + aimpoint_y**2)
        new_time_to_target = range / bullet_speed
        time_delta = new_time_to_target - time_to_target
        n += 1

aimpoint(iters=5, target_x=0, target_y=100, target_vel_x=1, target_vel_y=0, bullet_speed=100)