#Self-made Fly Control VS self-made cannon control and self-made Anti-air Shells
1 messages · Page 1 of 1 (latest)
I don’t understand how but I think this is really cool
bros in the thick of it right now 😭 🙏
Very sigmal
With create radars anyone will be able to do this
Once it gets a few updates
Which should be veeěęeery soon
did I do this too early or too late?
wdym
What feature would it be in create radar
auto targeting
but theres stuff missing like lead prediction and more settings for targeting
does it behaves like ship helms or it is changing the input to a wing bearing?
oh i meant the auto aaa
not auto plane controls
wait tell me how u did auto plane flying
we might do that
I used cimulink form control craft to build control algorithm and input to the bearing
it is a real control system
uhh what
there is a few purturbation in vs planes so PID is quite enough
whats cimulink form control craft?
control craft provides logic circuits
and you can link them and compile to build a circut
is that some mod?
an addon for cc?
for valkyrien skies
also provides interface to cc
but not necessary when using
also I wonder how you implement auto targeting for AA stuff, the function of shell trajectory is bit complex
yes we just calculate it
im still puzzled at how u made auto planes tho
all i understood was to use control craft
I used polynomial fit instead of iterators so it does not work for the enitre field defined by possible hit points
what are u talking abt rn? planes or aa?
that poly fit is for AA
for plane, sense the pose first, then input to a PID controller
and it generates a control to the bearing, so it got controller
controlled
for example, you need to roll +5 degs, and currently at -2 deg
oh u mean for leading the targets. yeah we plan to iterate it, but whats a polynomial fit, havent heard of it
yep ive heard of pid but dont know what it does exactly
poly fit is like, if you have a function that is very complex and cannot be expressed as a function that can directly calculate for result
then you can draw the figure first, and use polynomial to draw a cruve that close to it
so it is faking another function but you can calculate a similar result in a valid interval
man that sounds complex. but im guessing this would make the guns account for constant dodging, right?
sorry for not understand all the technical yap. but i am interested in understanding it
if you are asking about pid, you can basically understand like, p is proportional, i is integrate, d is derive
then you have 3 numbers from p, i and d of the error
multiply them with gains and add together
that is the input to the bearing
what about the poly fit stuff. i didnt quite understand it
that is a math technic
for example you have seen the function of cannon trajectory right?
when you have distant d, angle theta, height difference dy and set dy and d as known, you will find it not possible to directly calculate an angle based on what you have
then you need to use a iterative method to solve this like bisection
trajectory is easier than prediction
yes i know. we solved trajectory by looking at how others did it, which was some really complex math function that we copied lol
well polyfit is that you solve the funcion with an iterative method first, and draw them in a curve, then use a polynomial to have a function like y = 1 + x^1 + x^2 +x^3 + x^4 to instead the original function
then you don't need to iterate each time it calculates
it is less accurate but faster
the hardest point from my view is to calculate the correct time
did u just iterate angle by target velocity * time to target at x target position?
so far yes but that is the easiest one of methods
you can use a time gap to calculate for average velocity
also you can use cubic spline to predict the trail
but mostly time* vel is enough
i heard that before but no idea what its. also is time gap just sampling velocities over time?
also these methods is only one side of prediction, probability also works but that is outside of my knowledge
probability?
cubic spline is drawing a function of 3 orders instead of 1 order which is a straight line
that is another field in predicting stuff, for example, the AI you see is from probability
but that is outside of my knowledge, I just know there is a field
for time gap stuff ,it is like replacing euler's method with runge-kutta, well euler's is the easiest prediction as y(t+dt) = y(t) + y'(t)dt
I forgot the deriving progress from runge-kutta but the result is use the average velcocity of current tick and previous tick instead
ive heard of runge kutta but that was from orbital mechanics, and not sure what that was. also dont know eulers method
euler's method is just sounds hard, it is that method using time*velocity to predict
which is the method you are using I think
well, its the method we plan to use. ive only used it in unity before
anyway, if the time is short, euler's is enough
but time is kind of hard to calculate from cannon trajectory
for example, you don't know the time first so you cannot calculate for a prediction, so it is tend to start from the current position without predicting, but once you calculate the time and add for prediction, that distance from you to the hit point of shell changes
then there is a slight change in time