[Stormworks] I tried making MRAP! Equipped with an active protection system!

Рет қаралды 5,696

Mimi the Cat Lover

Күн бұрын

Пікірлер: 16

@chasibu-n3v Жыл бұрын

宇宙の法則が乱れてたのに相変わらず兵器作ってる…

@SakuraNeco-t8q Жыл бұрын

凄くいいですね！物理装甲いろいろ重いので代わりにこれ載せてみます

@Tagalyaga Жыл бұрын

I'm still failing to make a proper missile since the aerodynamics update. Good job on the APS, and thanks for explaining how it works !

@Chopstx420 Ай бұрын

hey ive tested this and when the missile goes below the aps shoots up

@Nekozuki_Mimi_Moka Ай бұрын

レーダーの精度が低いので、敵ミサイルとは全然違う方向にプロジェクタイルが飛んでくことがあります。（特に、ミサイルが少し旋回している場合） Due to the low accuracy of the radar, the projectile may fly in a completely different direction than the enemy missile (especially if the missile is turning a little). (I used Google translation)

@Chopstx420 29 күн бұрын

@ I’ve actually fixed this by changing the logic on the missile itself, it is more reliable now

@Chopstx420 29 күн бұрын

@ instead of going the other way it flies into the missile

@reilly2828 Жыл бұрын

How do you bypass the radar noise to accurately judge where the incoming missile is?

@Nekozuki_Mimi_Moka Жыл бұрын

私は、線形最小二乗法を使ってノイズフィルタリングをしてます。線形最小二乗法は、目標の座標データに一番それっぽい直線を引いてくれるやつです。目標の座標=at+bのaとbを求めてくれます。 a = 座標データと時間tの共分散　÷　座標データの分散 b = 座標データの平均値　-　a×時間tの平均値以下のような手順で、敵ミサイルなどの位置を計算してます。 ①目標の座標を計算します。 ②LUAのテーブルに、目標の座標と観測した時間をデータとして記録します。 ③数[tick]～数百[tick]分のデータに、線形最小二乗法を使います。線形最小二乗法を使うのは、目標の位置データのXYZ軸に対してです。例えば、目標の位置データのx軸に線形最小二乗法を使うと、以下のような式が得られます。目標のx座標x[m]、時間t[tick]、速度a[m/tick]、切片b[m] x = at + b この式のtに今の時間を代入すれば、目標の今の位置が正確に求められます。 LUAの中で1[tick]につき＋１される、タイマーのような変数を作っておいてください。「その時間」や「今の時間」は、その変数の値から調べてください。線形最小二乗法について、できる限り簡単に書かれているサイトを見つけるといいと思います。意外と式自体は簡単です。ストームワークスで使う分には、証明までは知っている必要はないと思います。でも、LUAについての知識も必要になると思います。最小二乗法について、FCRの動画の中でちょこっと扱ってます。 kzbin.info/www/bejne/laeliqOGiNOph8U

@Nekozuki_Mimi_Moka Жыл бұрын

I am using the linear least squares method for noise filtering. The linear least squares method draws a straight line that most closely resembles the target coordinate data. It will find a and b of the target coordinates = at+b. a = covariance of coordinate data and time t ÷ variance of coordinate data b = average value of coordinate data - average value of a × time t The following steps are used to calculate the positions of enemy missiles, etc. ① Calculate the target coordinates. ② Record the target coordinates and observation time as data in the LUA table. ③Use the linear least squares method for several [ticks] to several hundred [ticks] worth of data. The linear least squares method is used for the XYZ axes of the target position data. For example, if we use the linear least squares method on the x-axis of the target position data, we get the following formula: Target x-coordinate x[m], time t[tick], velocity a[m/tick], intercept b[m] x = at + b By substituting the current time for t in this formula, the current position of the target can be determined accurately. Please create a variable like a timer in LUA that increases by +1 for each [tick]. Check "that time" and "current time" from the value of that variable. I think it would be a good idea to find a site that explains the linear least squares method in the simplest possible way. The formula itself is surprisingly simple. I don't think you need to know the proof to use it in Stormworks. However, I think you will also need knowledge about LUA. I briefly discuss the least squares method in the FCR video. kzbin.info/www/bejne/laeliqOGiNOph8U (I used Google Translate)

@reilly2828 Жыл бұрын

@@Nekozuki_Mimi_Mokasince it’s linear do you have to disregard old data as if the target or manoeuvres or begins to turn the data is wrong?

@Nekozuki_Mimi_Moka Жыл бұрын

相手が旋回している場合は、古いデータを無視しないといけないと思います。ですが、私もどれくらいのデータを無視すればいいのかわからないです。私が作ったFCRなどは、そういった対策は一切してないです。 If the opponent is circling, I think we have to ignore old data. However, I don't know how much data I should ignore. Therefore, the FCR etc. that I created do not have this countermeasure. (I used Google Translate)

@reilly2828 Жыл бұрын

@@Nekozuki_Mimi_Moka thank you for your help as so far you are the only person I know to use this filtering system which seems to be the best one available. (Also thank you for translating it for me, would it help if I translated my comments into Japanese next time?)