Monday, March 21, 2011

XY Velocity for a projectile? Solved for

Alright, So last I left off I had coded how to deal with projectiles moving in a strictly horizontal or vertical way. And was waiting on making the code for dealing with both velocities. At first I thought I would have to deal with some type of equation of two variables and solve form both of them using some complex business, but it turns out the method I use is at least 3x simpler than that.
So, first off I use the same method I've been using to get the time interval I should be using in my estimates. since I've described that in previous blogs I won't bother redefining it here. From that I get my hBound variable which is the top of said time interval. From here, I can compute how far the projectile will travel in the time given that it travels at a constant speed of 2.0 units. I then figure out the heading of my turret to aim at so it points at the targets estimated position after hBound time. From here all I need is the vector to give to the projectile as its x movement and y movement. So, using the cos of the gun's theta times the distance of the projectile as the x component and the sin of the guns theta times the distance of the projectile as the y component, I make up my vector. From this I have a triangle that may or may not overshoot the position if fired at the wrong time. So how do I figure my delay time out?

Simple, I take the magnitude of that vector I just made and divide by hBound to get how many units of time it will take to traverse this hypotenuse I've created. If that number is equal to the time it takes for the projectile to get to the estimated position then hoorah fire right away, weapons free, danger close. If its less than that then set up delay which is as simple as Time of the target - Time of the projectile. And the other situation, where T(proj) > T(target) never occurs because that would violate the work above with the whole hBound thing, theres a reason it's called the hBound you know. From here things get pretty simple.

since I have my heading, my delay time (if there is one) all that is left to do is to fire. Which is done by created  anew object that has the position of the turret, the velocity vector is the heading vector of the turret(which has been normalized) multiplied by the velocitiy I know the projectile will travel (2.0 units), ahem. Tah-dah.

Now of course is where I fork myself, should I proceed to trying to create a 3d version of this? Or should I build up a simple library in DirectX to actually see whats going on instead of just numbers? Hmmm.

Sunday, March 13, 2011

Horizontal Target Tracking

Woo!

Just finished coding the turret of my current project to lead a target thats moving only horizontally. It took me a while to wrap my head around alot of the numbers, and also I was accidentally using degrees to start off my turrets heading and not radians. Wups!

Anyway, so this is how I accomplish leading a target horizontally:


 first using vector arithmatic and dot products and arcCos, we get the heading we'll be firing on.
 then we account for possible time delays in order to account for differences in distance.

 Heres the code: there is some stuff missing that is declared a bit above, such as getting the hBound which is the time where the distance the projectile will travel becomes greater (or equal to) than the distance the target will travel in the same amount of time. This is used so that we can lead the target. But because we lead it, we need to make sure to fire at the correct time in order to score a direct hit. So thats why we do some subtraction between the amount of distance and time and all that. Anyway: Heres the code:



if(target.velocity.y == 0 && target.velocity.x != 0){ //target only moves horizontally
                //get the heading to fire on.
                estMagPos = estMagPos + estMagVel*hBound;
                Vector2D gt =  (estMagPos - gun.position ).Normalize();
                gun.theta -= acos(Dot(gt,gun.heading));
                gun.heading = gt;
                //now, account for leading and time delay
                //hBound * vel = how much distance the proj will travel. 
                //Use the gun.theta and this to resolve the vector into it's components
                //then the distance between the (hBound * vel) - x of target
                double projd = hBound * vel;
                projd = (cos(gun.theta ))*projd; // x distance of vector
                double deltax = projd - estMagPos.x;
                double waitT = deltax/vel;
                gun.countdown = -waitT; //negation to make positive. no neg times
                gun.locked = true;
                std::cout << "Target Locked. Countdown Til Firing: " << gun.countdown << std::endl;
                system("hal target lockd");


So, the calls to system and hal are just a speech program to say the string aloud. 

Saturday, March 12, 2011

Getting a bit farther with my vectors

 So I've gotten a bit father with things, these two images show some of the thought process on what I'm working on right now.
Which is, targeting and leading when the object is moving horizontally only. I've already worked out the case for the stationary target, which is the easiest, so now I'm working on the next easiest (I think) which is just horizontal movement.
this image right here shows how I was trying to think about zones and ranges of the turret.


normally, I would go into detail on more stuff. But I'm pressed for time right now, but once I get back to the world of internets, then I will gladly post and describe all the difficulties I've been having and how I've been solving them.

coming close of vacation.

So, my vacation is starting to come to a close. And I'm looking at what I expected from this vacation, and what I actually got.

I expected that I would, maybe not get laid, but find someone here who might be a friend who I could share attraction with. And maybe have a brief fling. I did kiss one of my friends, but it's normal for her and I to kiss. We're close, and the kissing really doesn't mean much. I kiss a few of my female friends on the cheeks when I see or leave them, this one I kiss on either the cheek or the lips. It's not a romantic kiss though, it's kind of like an italian kiss I suppose because it's more friendly than anything else. Anyway, getting off topic here aren't I? I expected to maybe cuddle with a specific friend and to admit some feelings for her. That didn't happen, because even though I saw her twice... maybe three times, it was always with a group. And since she invited people to events that we had planned on doing. I'm taking her hint that she just wants to be friends who tell each other they love each other and act like boyfriend girlfirend without the actual physical affection and only the virtual. It's just fake. I'm not a fake person, most of the time, and I don't especially enjoy fake. I do put on my smile for a good amount of the day, and I tend to shrug off my insecurities and sweep all my problems under an apathetic bed. I don't want a fake relationship, and I don't want a fake love, because after pretending for so long, it starts to feel like it might actually be something. I'm tired of that illusion. ... cough. I expected to see lots of friends and catch up on old times. That kind of happened, with one of my friends. The weird thing is, it was a friend who I am close with, but she was the only one who actually gave a damn to ask about me. Most of my conversations go something like this:
Hi, hi, how are you, and then they go on for a while, when they stop, they don't ask about me, so I further the conversation by asking them how their lives are going. And then the topic is them, them, them. I wish that for once or twice, maybe I could be the topic.

I think I'm mainly annoyed at the level of drama that occured while I was home. As noted by my previous rather depressing blogpost, my friend screwed up his relationship with his girl, and happened to do it with the girl who I was falling for. It's the reason I took a walk yesterday. Loaded the playlist I talked about onto my mp3 player and went for a walk in the crappy overcast weather. It was good, I got a good amount of thinking done. I was trying to figure out if I actually really like the girl enough to go for it, or if I'm more attracted to her as a friend. I didn't really figure too much out. But, I do think I'm attracted to her based on our common fuck ups. A'la she has baggage. Much in the same way I do. She still thinks about her big ex, and mentions him a lot. He was a big part of her life, so its no wonder a lot of things still remind her of him. And I think I like that about her, because I feel the same way about my ex from a long time ago. Everything still reminds me of her. Maybe that's why coming back home is kind of painful sometimes.

I expected more from my vacation, maybe some sunshine. I know I can't control the weather, but I've seen the sun once this entire week. It's frustrating when you want to go out and do fun things with friends, but they dont want to because the weather is shitty. It's rather annoyying and difficult.

I guess, when it comes down to it, I am looking forward to go back to VT, when I've talked to people over SMS, I've had to erase a part of a text that said: I only have x days before I go back home, multiple times, I guess I consider VT more home than this place to some extent. Maybe it's just nice to have a freshstart and not have all the people who already know you and stuff. Start new, find someone new, who hopefully wont fuck up my mind like my ex did.

Wednesday, March 9, 2011

Turret Estimation

Workin on my program some more:

So I've come up with a bit more code, and a few more interesting problems. I think I've figured out how to use my vectors correctly. I can use velocity vectors and position vectors to estimate the locations of objects. It's great.

The problem I've come up with is how to figure out where to rotate my turret to in order to fire. Because I have a few variables I need to keep track of, and keep in mind.

1. The projectiles statistics: the bullet or shell to be fired by the turret has a velocity vector and position vector. Once I move into a 3d space I may include a mass in here. But besides that, velocity is a straight float, why? Because I don't know which x y components to give the vector because I don't know the theta. so, a straight number like 2.5 m/s or something, allows me to estimate where to shoot the bullet. This will make sense once I describe my idea for this.
2. The target objects estimated path using the pings. This gives me a velocity vector and a position vector for the target.
3. The current heading, or theta of the turret.

So, my method, is briefly describe in the picture below.

I ran atrial of it last night, and it appeared to be working, but I do need to iron out some kinks in the simulation

Thursday, March 3, 2011

Radar Tracking Turret!

I came up with a new idea to code! It will probably take up a good chunk of my nights during spring break

I simple turret that tracks and fires at objects in its pinging field, in 2d space. Once I get that working, I'll think about working in 3d space.

Here are a couple of my sketches for ideas:










ok, so the first image is how the ping system will work. The second is how the pings data will be used to determine the vector of the object being targeted. and the last is how the turret will estimate where to fire.

So this is how I plan to do it. No code yet!
Have the turret object which will hold its own cordinates in its plane, its angle of firing and a pointer to its projectile loaded.
The projectile object will have a velocity and x,y cordinates. (once I move to 3d space it will have mass and other things)
The target will have an x,y and a velocity and theta for movement.
Then there is the collections object that will store all these things. this will be used by the pings sent out by the turret to let the turret know the location of the target object.

A few of the details still need to be worked out. But I've got enough that I can start a basic console prototype

Wednesday, March 2, 2011

OCaml Pattern Matching and etc.

I've a midterm tomorrow, in 3 classes. Anyway, calculus isn't hard. German will be alright. But programming languages, not that it will be hard, but just.... Ok so, we're allowed to use our notes we've taken in class as well as the daily summaries that he has given us in the past. Unfortunately, I've been taking all my notes in LaTeX on my computer. Which means I need to print them. Which means I need a printer, which means I needed to get my roommate to print it out. Which means I went and did something fun and asked him to do it. I came back a while later, and bam. Thirty Five Pages. THIRTY FIVE BLOODY PAGES of notes. Granted that's for the entire first half of this semester. But... It's computer paper! Formatted! To what looks like 10 size font. That's a lot of characters. Plenty of bits.

Anyway, besides that, I've just completed programming a reverse polish notation calculator in OCaml. The powers of pattern matching are spectacular. For anyone who has never programmed in Objective OCaml, I'd recommend attempting to. The type system is very nice, even if trying to figure out your error messages is a pain in the ass. I normally love my C++ but I do enjoy OCaml's elegance factor.

This semester seems like it's been quite a jump in what I've learned as far as CS goes. Before this semester I had no idea how to use php, javascript, server side includes, ocaml, unix file commands, batch file commands, and some directX code. Its a trend I would like to keep, this upward mobility of my learning. I like.