Possible additions to the event block and the offensive AI block

Antipindoc shared this feedback ago

Not Enough Votes

A possible addition to the event in the event block are locked target distance. This can be useful for missile systems, e.g. a missile releases a bunch of independent warheads when its within a certain distance. It can also be useful when considering the offensive and defensive AI Blocks. For example: A mining drone with guns onboard will always flee, if an enemy is approaching, but if its too close and combat cant be avoided, then it will switch to the offensive block.

My second suggestions is about the offensive AI block. First, some tooltips should be added to the guidance types, since their functionality can differ depending on the task. Secondly, when testing the "intercept" mode, I noticed that the AI can have trouble hitting its target, even if the target is stationary and the AI has thrusters available in all directions. The situation gets quite worse, when trying to use a single forward thruster. So I used a 1x1x8 grid that is inspired by a real life Air-to-Air missiles. This design can only have a single thruster and I used the "Target prediction" mode expecting the missile to automatically adjust its orientation parallelish to the target to be able to catch up and/or properly adjust to deviations. The missile refused to do that. I noticed that the "intercept" mode relies heavily on all-directional thrusters, instead of actively using gyroscopes to correct its course, no matter if it was in zero gravity or not.

If u have any comprehension questions regarding my suggestion, I'd be happy to answer them.

I like this feedback

Replies (1)

1llusionist ●

have you tried the proportional mode? it seems to fly better with designs that have less thruster vectors available, but lacks a bit more in terminal guidance compared to target prediction. if your design is repsonsive enough a sensor to switch modes might help, but so far i think it needs to switch a little earlier than the sensor can detect.

Reply URL

Antipindoc ●

yes, Ive tried all of them, but target prediction seems to be the most reliable when hitting stationary and moving targets. Ive tested a bit and noticed that for some reason basic and target prediction utilises only its all-directional thrusters until its about 200m within target range and then starts correction with gyroscopes. I wish that the grid would always try to point its nose on target since launch, because that would semi-fix the problem.

URL