nikolas93TS -> RE: more independent unit AI (6/11/2020 12:16:40 AM)
|
Sergey Suvorov's volumes on the BMP-1 and BMP-2 mentions some interesting tests regarding auto-cannons and tanks. In one attempt to persuade military officials to adopt the Object 681 with the 2A41 73mm smoothbore cannon, a comparison trial was organized. The target was a T-72 tank, and the distance was 1,200 meters. The Object 681 opened fire first (on the side of the tank), and not one single round penetrated the tank. One went over, one fell short, and the other successfully impacted the side skirt, but the reinforced plastic side skirt flexed in such a way that the round did nothing. Then, the Object 675 - the BMP-2 prototype - stepped up. It let off three bursts of 8 rounds in the high rate of fire mode. All external equipment including periscopes and the gun-sights were completely destroyed, and the roof mounted anti-aircraft machine gun was separated from its mount, landing 15 m away from the tank. Later on during its acceptance trials, the BMP-2 fired on a combat loaded T-55 (modernized) and T-72 from the side, again with 3 x 8 round bursts on the high rate of fire mode. In both cases result was cleat mission kill. The T-55 had its anti-aircraft mount, IR searchlight, and externally mounted laser rangefinder shredded away. The 100mm cannon was hit and penetrated in two places. The external fuel tanks mounted on the track fenders ignited and burned outside the tank, but the gun-sight was unharmed and tank was in a drivable state. A total firepower kill was achieved due to the damaged cannon. The T-72 also had its anti-aircraft machine gun ripped off, and its optical devices were damaged. The fender fuel tanks were hit and the ignited damaged the turret seals (compromised the sealing of the NBC system) and some of the burning fuel leaked to engine compartment and damaged the engine, though the automatic fire suppression system worked. It was counted as a mobility and firepower kill. In neither case was side armor actually perforated. Currently auto-cannons in-game are pathetic against tanks (although lucky shoots of more powerful cannons can still disable them in certain cases, and in actual combat above situations are probably less common) but we are planning a more detailed component damage model. Still, I just wanted to illustrate how hard is to judge and even more so to code what and when AI should stop firing. Sometimes the player might want to suppress, damage, or distract the target it another unit can kill it. There were some recent AI self-preserve suggestions too, but we haven't really done anything yet in that field.
|
|
|
|