APN Guidance - From the Depths

Рет қаралды 1,574

Bungalowbill

Күн бұрын

A quick tutorial about how APN guidance works and how to know if it is tuned properly.

Пікірлер: 22

@dreekfire721 2 жыл бұрын

Hey, great video! I think APN was always one of those mechanics that everybody uses but nobody understood. I implemented it for a lua missile so I can offer some input: 1. Slight nitpick, you said the missile tries to maintain the target at a constant bearing, for example 15 degrees to starboard. It's actually not relative bearing, but absolute bearing, so it's more like it tries to maintain the target directly north, for example. 2. The "augmented" in APN refers to gravity compensation. Real missiles are affected by gravity whereas their targets are assumed to have lift which counteracts gravity, so the augmenting term makes the missile aim high to avoid getting pulled below the target by gravity. This is a lie in FtD because FtD missiles are unaffected by gravity as long as their engines are on, so it would more accurately just be called PN. 3. I found it interesting that when you set the gain really low, it behaved like direct guidance. Perhaps this is what the devs actually meant by "augmented" but if that's the case, it's an incorrect use of the terminology.

@bungalowbill521 2 жыл бұрын

The worst part is that I already knew it was absolute bearing because I was just tinkering with my breadboard and I still got it wrong. I almost never make it through one of these videos without some sort of mistake.

@asrieldreemurr5029 2 жыл бұрын

All kinds of proportional navigation try to keep relative bearing constant: absolute bearing remains constant too in the case of non-accelerating targets, but is otherwise irrelevant. APN is called "augmented" because it takes into consideration target acceleration (derivative of the rate of change of relative bearing) and not just speed (rate of change of relative bearing): relative acceleration due to gravity is only one part of this term.

@dreekfire721 2 жыл бұрын

Oh yeah, my bad. I originally saw a derivation of APN assuming a non-accelerating target, so the acceleration term only included gravity, and I forgot that it should also include the target's other acceleration. However, it is not the second derivative of the LoS, but rather just the target's acceleration vector. Also, it is change in absolute bearing which determines the response, not relative bearing. This can be seen in the definition of PN, and can be derived from the zero-effort miss (ZEM) formulation. In the case of a missile already on the correct course to a non-accelerating target, neither absolute nor relative bearing change. Whether either of these is "relevant" depends on what the question is. My source is "Modern Missile Guidance" by Yanushevsky, the wikipedia page on PN, and various online lectures and research papers about more modern variants of PN which I can try to dig up if you're interested.

@asrieldreemurr5029 2 жыл бұрын

@@dreekfire721 I would suggest reading through your sources again: the wikipedia page, in particular, makes it fairly obvious that absolute bearing is not relevant by expressing acceleration in terms of line of sight. Also, measuring LOS and its derivatives is easy, measuring actual target acceleration is difficult and prone to errors: the simplicity of PN is one of the main reasons behind its success.

@dreekfire721 2 жыл бұрын

@@asrieldreemurr5029 I've read through these sources multiple times, and I've read through multiple sources. They all corroborate in that it's based on absolute bearing, not relative. Since you said so, I went back and read the Wikipedia page one more time. It defines the rotation vector of the LoS as RxV/R*R, which is the rotation vector of the absolute bearing, not relative, and uses this vector to define the commanded acceleration. Quoting from Yanushevsky, which the wikipedia page cites, "the LOS direction relative to the inertial coordinate system is kept constant." Since it is relative to the inertial coordinate system, this is absolute, not relative bearing. Additionally, the mathematical justification for PN guidance is that it is merely a different way of writing the acceleration needed to cancel out the zero-effort miss. This derivation only works with absolute bearing. Finally, relative bearing would not work for a vehicle with multidirectional thrusters, particularly one in space where there is no atmosphere to define a fixed reference frame so it's impossible to determine a missile's own bearing. I agree that measuring actual acceleration is harder than measuring the second derivative of the line of sight. Checking again, these two can both be used for the same purpose; the second derivative of the LoS can be used to estimate the target acceleration perpendicular to the LoS. So in the case of True PN, you can use either, although Pure PN requires more acceleration data.

@GMODISM Жыл бұрын

Very informative, I had them high but someone recommended having them really low so just wanted to see a nice comparison :)

@rabidr3aper213 2 жыл бұрын

This is a great video. You should make one on meta armor schemes. A video on vectored thrust would be very welcome as well.

@a1b2c4d1 2 жыл бұрын

Dude this was awesome. I was not fully sure how adjusting the Gain worked and if PIDs taught me anything it's that high gain sliders were a BIG no no. I'll definitely be seeing more connections and be using 1-2 fins now instead of 3-4. Great vid!

@thecanadiankiwibirb4512 2 жыл бұрын

Most of my craft use hydrofoils on a misc axis connected to a high gain PID to maintain an extremely constant altitude. So I think there is a place for them

@bungalowbill521 2 жыл бұрын

"High" is always relative. The rule of thumb for gain anywhere is to turn it up until it starts to wiggle, then turn it down until it just barely stops.