Hello TL!

For my 4th year (last year here in Quebec) I'm working on a big project called "Capstone" which I think most north american universities require their eng students to complete before graduation.

Since I'm in Mech, I'm actually going to BUILD this thing, but as a good little engineer I need to prove to my supervising professor I'm not an ass and actually know this thing can work. So, instead of using SolidWorks or working model, I just programmed in straight in C++/Irrlicht!

Here are some screens:

Here is the executable:

Capstone quadrotor simulator project

Please test it out, the readme should be plenty helpful.

Let me know if you see any issues with the physics or controls, and if you have any graphical advice. My knowledge of sciences in order of magnitude goes PHYSICS+MATH>>>>>>>programming>>>>>>>>graphics so go easy on me

Thanks!

PS: let me know what your FPS is! if it's below 100, please provide PC specs

frogmelter

United States971 Posts

October 11 2012 17:04 GMT

Weird chrome is providing a suspicious file alert....

RoyGBiv_13

United States1275 Posts

October 11 2012 18:33 GMT

Nice work! 3d engine with textures and collision detection! Being able to program while also do the mechanical design makes you a prime candidate for modern engineering jobs. Did you open source this?

CoughingHydra

177 Posts

October 11 2012 18:57 GMT

How do you detect collisions between objects?

Recognizable

Netherlands1552 Posts

October 11 2012 19:14 GMT

Wait, for your final year in univeristy you are going to build that cross chopper thingy? Kids in my class want to do that for their exam year, in high school. I already expected them to fail hard, but now I don't even think they stand a chance haha.

Abraxas514

Canada475 Posts

October 11 2012 23:19 GMT

On October 12 2012 03:33 RoyGBiv_13 wrote:
Nice work! 3d engine with textures and collision detection! Being able to program while also do the mechanical design makes you a prime candidate for modern engineering jobs. Did you open source this?

Didn't open source yet because it's part of my capstone project and I need to keep it as such until the project is complete. Thanks for the comment!

On October 12 2012 03:57 CoughingHydra wrote:
How do you detect collisions between objects?

The following technique works best. You save an array of all the tris of all the models you want to test collision with. You then perform a direction + velocity test against each tri in "collision distance". If your line intersects, you use those parameters to determine the output. I just directly subtracted a fraction of the velocity dot producted onto the normal of the triangle.

On October 12 2012 04:14 Recognizable wrote:
Wait, for your final year in univeristy you are going to build that cross chopper thingy? Kids in my class want to do that for their exam year, in high school. I already expected them to fail hard, but now I don't even think they stand a chance haha.

It's a quadrotor. And there is no way a highschooler can build this without using a flight controller, because the control sequences need physics analysis to compute. You need to have studied Statics, Dynamics, Mechanics of Materials, Control systems 1 (at least), Fluid dynamics and also all the math: differential equations, vector calculus and transform calculus... to program the flight controller.

It sounds like they want to buy one of these:
http://www.hobbyking.com/hobbyking/store/__592__501__Multi_Rotors_Parts-RTF_PNF_ARF.html
and put it together. Good for them, every engineer needs to start somewhere.

BTW if one of the DLLs are giving you errors, just delete it. I included them for redundancy but forgot to give instructions.

Ssin

United States88 Posts

October 12 2012 03:56 GMT

This is very cool, and I will be sure to play around with it. Physics and math are really fun, and when you mix it with programming you can do all kinds of interesting experiments.

Hope it all works out for you!

DarkwindHK

Hong Kong343 Posts

October 12 2012 04:05 GMT

man this is really a great work. I study physics in my bachelor degree and I really do not know any programming; just the math and physics you need to know to program will take years to master!

With 4 rotors there is a lot of calculation going on and they all affect each other and the angle of the Quadrotor itself......

Abraxas514

Canada475 Posts

October 13 2012 17:18 GMT

On October 12 2012 13:05 DarkwindHK wrote:
man this is really a great work. I study physics in my bachelor degree and I really do not know any programming; just the math and physics you need to know to program will take years to master!

With 4 rotors there is a lot of calculation going on and they all affect each other and the angle of the Quadrotor itself......

Here is an excerpt from my control systems code, well commented as a mark of a good programmer

//***** FORCE AND ANGLE OUTPUT CONTROL*****//

// Here is the actual control system.
// First we need to figure out how much force in the Y do we need to balance:
f32 yforce[5] = {0, (vbalance + vycorrecttion * .5), (vbalance + vycorrecttion * .5), (vbalance + vycorrecttion * .5), (vbalance + vycorrecttion * .5)} ;

//Next we adjust yforces for losses while rolling and pitching:
f32 lossfactor = abs(cos(plane.theta.X

EGTORAD)) * abs(cos(plane.theta.Z

EGTORAD));

yforce[1] /= lossfactor; yforce[2] /= lossfactor;
yforce[3] /= lossfactor; yforce[4] /= lossfactor;

// Next we add/subtract some force to produce roll
yforce[1] -= rollcorrection * rollcf ; yforce[3] -= rollcorrection* rollcf ;
yforce[2] += rollcorrection * rollcf ; yforce[4] += rollcorrection* rollcf ;

// Next we add/subtract some force to produce pitch
yforce[1] += pitchcorrection * pitchcf; yforce[2] += pitchcorrection * pitchcf;
yforce[3] -= pitchcorrection * pitchcf; yforce[4] -= pitchcorrection * pitchcf;

// Next we change diagonal forces to produce yaw
yforce[1] += yawcorrection * yawcf; yforce[4] += yawcorrection * yawcf;
yforce[2] -= yawcorrection * yawcf; yforce[3] -= yawcorrection * yawcf;

// Next we figure out how big our actual force output must be:
f32 totalforce[5] = {0, yforce[1] / cos(plane.f[1].theta), yforce[2] / cos(plane.f[2].theta),yforce[3] / cos(plane.f[3].theta),yforce[4] / cos(plane.f[4].theta)};

//// Increase by speedfactor:
totalforce[1] *= speedfactor; totalforce[2] *= speedfactor;totalforce[3] *= speedfactor;totalforce[4] *= speedfactor;
//
//// Recalculate y:
yforce[1] = totalforce[1]*cos(plane.f[1].theta); yforce[2] = totalforce[2]*cos(plane.f[2].theta);
yforce[3] = totalforce[3]*cos(plane.f[3].theta); yforce[4] = totalforce[4]*cos(plane.f[4].theta);

// Next we assign the X-component:
f32 xforce[5] = {0, totalforce[1] * sin(plane.f[1].theta), totalforce[2] * sin(plane.f[2].theta), totalforce[3] * sin(plane.f[3].theta), totalforce[4] * sin(plane.f[4].theta)};

// Next Assign Z component:

f32 zforce[5] = {0, 0, 0, 0, 0};