Re: [Flightgear-devel] Yasim static friction?
On 07/05/2012 02:41 PM, Viktor Radnai wrote: Thanks for that! So just to clarify -- this is a bug in Yasim code (or more like a missing feature) and I'm welcome to fix it?:) I'm just an absentee hacker, so I can't say what is or isn't acceptable any more. But it seems like a sane enhancement to me. But broadly yes: it's free software. The whole point is to make it do what you want. Andy -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Yasim static friction?
Hi there,
The rest of this mail will just deal with the gear friction issue for now.
Having looked at Gear.cpp I think I can understand the issue a bit
better. The method that calculates tire friction is Gear::calcFriction.
It looks like this:
float Gear::calcFriction(float wgt, float v) //used on solid ground
{
// How slow is stopped? 10 cm/second?
const float STOP = 0.1f;
const float iSTOP = 1.0f/STOP;
v = Math::abs(v);
if(v STOP) return v*iSTOP * wgt * _sfric;
else return wgt * _dfric;
}
Note the constant in there that marks the point where static friction
starts to get used. Also note that the constant is multiplied by v, so
at zero speed, there is zero friction. This is perfectly fine on water,
but not on solid ground. It also explains what's happening -- the force
of the wind pushing against the plane will meet with no resistance until
it starts to equal v*iSTOP * wgt * _sfric and the plane will eventually
roll with that constant speed.
The reality is that on solid ground the force caused by static friction
will be equal to the force acting upon the wheel (the bearings
actually), up to a threshold of Fmax, when the wheel/bearing assembly
will unstick and start to rotate.
Right now it seems that to fix this, reality must be modelled in this
respect. If I take Andy's suggestion and just try to clamp the static
friction to an arbitary fixed value, then the force returned may or may
not be larger than the force pushing against the wheel, and the result
will be that the plane rocks with the (3 knot) wind while very slowly
turning into the wind, while stopped on tarmac...
The question is, where do I get the force acting upon the wheel?
To try to figure that out, I had to look into Model.cpp where I ran into
further difficulties:
// The landing gear
for(i=0; i_gears.size(); i++) {
float force[3], contact[3];
Gear* g = (Gear*)_gears.get(i);
g-calcForce(_body, s, lv, lrot);
g-getForce(force, contact);
_body.addForce(contact, force);
}
Here the forces created by fricion on all the gears are added up. So if
I was to keep the model stationary without any jitter, the sum of the
resistive forces from the gears must exactly equal the sum of the force
acting upon the gears. This is out of scope of just modifying Gear.cpp
and definitely out of my league for both the knowledge of C++ and the
Flightgear codebase.
Could someone please help me with this? Thanks in advance!
Cheers,
Vik
On 07/05/2012 09:34 PM, Andy Ross wrote:
(Happened to be browsing in time to see a question)
On 07/05/2012 06:21 AM, Viktor Radnai wrote:
I don't see any obvious properties to set to take the engine's
resistance to turning over, or the friction of the wheels into account
to stop these unrealistic things from happening. How should I go about
fixing them?
That sounds right to me. Aircraft parked in gentle winds weren't really
part of the original test regime. :)
For the gear thing, see Gear.cpp:450 or so, and look at clamping the
static friction coefficient to some minimum value (probably tunable).
For the engine, you can likewise add some fixed negative torque value
near PistonEngine.cpp:214 to model internal resistance (currently the
code only models output power).
Andy
--
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel
--
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Yasim static friction?
Well, I am giving it a shot, but it seems it won't be that easy. See my other (long) mail sent on this topic :) Cheers, Vik On 07/06/2012 05:48 PM, Andy Ross wrote: On 07/05/2012 02:41 PM, Viktor Radnai wrote: Thanks for that! So just to clarify -- this is a bug in Yasim code (or more like a missing feature) and I'm welcome to fix it?:) I'm just an absentee hacker, so I can't say what is or isn't acceptable any more. But it seems like a sane enhancement to me. But broadly yes: it's free software. The whole point is to make it do what you want. Andy -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Yasim static friction?
On Thursday 05 July 2012 15:21:24 Viktor Radnai wrote: 1. When the aircraft is parked with no parking brake, it will usually start to roll slowly backwards -- pushed by the wind and maybe the runway slope. If I start the engine on idle, the thrust generated by the idle prop might stop this roll. On tarmac, even a 3 knot wind is enough to start pushing the plane back. On grass more is needed -- maybe 20 knots? Hi, In that case, even in real life, there's no other friction at play than the friction inside the wheel bearings, friction which is very low, almost ignorable. Regards, Emilian -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Yasim static friction?
Technically, you also have the compression of the wheels -- when the wheel turns, part of the tire in the front gets compressed while the back gets uncompressed. This is also a friction loss. Plus there is some friction between the rubber and the road. But my point is that the amount of friction is wrong. The plane should not start rolling with a 3 knot wind, that's nonsense. Static friction is greater than dynamic friction. If you stop your car on moderately flat terrain with no brakes, it will not start to roll easily, not even if it's windy. If the terrain is flat, it will stay put. If the slope is greater, it might start to roll very slowly with the bearings turning in stick/slip mode and then finally it will just roll and accelerate. So the plane should stay put until there's a 20-30 knot wind blowing against it. Probably more on grass. On grass with the real plane you might need half throttle or more to get it moving. The funny thing is that this part seems to be modelled correctly in Flightgear, so no idea what's wrong with the effect of the wind :) Same goes for the prop, the resistance the shut down engine offers is too small and so it's turned over way too easily. Turning the prop of the Falke at about 2/3 of its span requires about the same amount of force as lifting an object that weights 1-2 kg. This is a rough guess and the actual force varies during the compression cycle but you get the idea. For the wheels, I can try to make sure that the brakes are always slightly set, but what do I do for the prop? Cheers, Vik On 07/05/2012 03:30 PM, Emilian Huminiuc wrote: On Thursday 05 July 2012 15:21:24 Viktor Radnai wrote: 1. When the aircraft is parked with no parking brake, it will usually start to roll slowly backwards -- pushed by the wind and maybe the runway slope. If I start the engine on idle, the thrust generated by the idle prop might stop this roll. On tarmac, even a 3 knot wind is enough to start pushing the plane back. On grass more is needed -- maybe 20 knots? Hi, In that case, even in real life, there's no other friction at play than the friction inside the wheel bearings, friction which is very low, almost ignorable. Regards, Emilian -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Yasim static friction?
Does this happen only left wind, right? Or happens on right wind too? Hyde (2012年07月05日 10:02), Viktor Radnai wrote: Technically, you also have the compression of the wheels -- when the wheel turns, part of the tire in the front gets compressed while the back gets uncompressed. This is also a friction loss. Plus there is some friction between the rubber and the road. But my point is that the amount of friction is wrong. The plane should not start rolling with a 3 knot wind, that's nonsense. Static friction is greater than dynamic friction. If you stop your car on moderately flat terrain with no brakes, it will not start to roll easily, not even if it's windy. If the terrain is flat, it will stay put. If the slope is greater, it might start to roll very slowly with the bearings turning in stick/slip mode and then finally it will just roll and accelerate. So the plane should stay put until there's a 20-30 knot wind blowing against it. Probably more on grass. On grass with the real plane you might need half throttle or more to get it moving. The funny thing is that this part seems to be modelled correctly in Flightgear, so no idea what's wrong with the effect of the wind :) Same goes for the prop, the resistance the shut down engine offers is too small and so it's turned over way too easily. Turning the prop of the Falke at about 2/3 of its span requires about the same amount of force as lifting an object that weights 1-2 kg. This is a rough guess and the actual force varies during the compression cycle but you get the idea. For the wheels, I can try to make sure that the brakes are always slightly set, but what do I do for the prop? Cheers, Vik On 07/05/2012 03:30 PM, Emilian Huminiuc wrote: On Thursday 05 July 2012 15:21:24 Viktor Radnai wrote: 1. When the aircraft is parked with no parking brake, it will usually start to roll slowly backwards -- pushed by the wind and maybe the runway slope. If I start the engine on idle, the thrust generated by the idle prop might stop this roll. On tarmac, even a 3 knot wind is enough to start pushing the plane back. On grass more is needed -- maybe 20 knots? Hi, In that case, even in real life, there's no other friction at play than the friction inside the wheel bearings, friction which is very low, almost ignorable. Regards, Emilian -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel -- ** Hyde Yamakawa 308 Brookewood Dr. Peachtree City, GA 30269 Phone (770)632-6461 Cell (404)353-8758 e-mail: h...@hyde-tech.com http://www.hyde-tech.com/ ** -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Yasim static friction?
You mean when the wind is blowing from the front-left of the plane? Cheers, Vik On 07/05/2012 04:14 PM, Hyde Yamakawa wrote: Does this happen only left wind, right? Or happens on right wind too? Hyde (2012年07月05日 10:02), Viktor Radnai wrote: Technically, you also have the compression of the wheels -- when the wheel turns, part of the tire in the front gets compressed while the back gets uncompressed. This is also a friction loss. Plus there is some friction between the rubber and the road. But my point is that the amount of friction is wrong. The plane should not start rolling with a 3 knot wind, that's nonsense. Static friction is greater than dynamic friction. If you stop your car on moderately flat terrain with no brakes, it will not start to roll easily, not even if it's windy. If the terrain is flat, it will stay put. If the slope is greater, it might start to roll very slowly with the bearings turning in stick/slip mode and then finally it will just roll and accelerate. So the plane should stay put until there's a 20-30 knot wind blowing against it. Probably more on grass. On grass with the real plane you might need half throttle or more to get it moving. The funny thing is that this part seems to be modelled correctly in Flightgear, so no idea what's wrong with the effect of the wind :) Same goes for the prop, the resistance the shut down engine offers is too small and so it's turned over way too easily. Turning the prop of the Falke at about 2/3 of its span requires about the same amount of force as lifting an object that weights 1-2 kg. This is a rough guess and the actual force varies during the compression cycle but you get the idea. For the wheels, I can try to make sure that the brakes are always slightly set, but what do I do for the prop? Cheers, Vik On 07/05/2012 03:30 PM, Emilian Huminiuc wrote: On Thursday 05 July 2012 15:21:24 Viktor Radnai wrote: 1. When the aircraft is parked with no parking brake, it will usually start to roll slowly backwards -- pushed by the wind and maybe the runway slope. If I start the engine on idle, the thrust generated by the idle prop might stop this roll. On tarmac, even a 3 knot wind is enough to start pushing the plane back. On grass more is needed -- maybe 20 knots? Hi, In that case, even in real life, there's no other friction at play than the friction inside the wheel bearings, friction which is very low, almost ignorable. Regards, Emilian -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Yasim static friction?
The plane gets pushed either forward or backwards by the wind, depending on the wind direction. The prop seems to only get turned in the direction of normal rotation, even in a tailwind, which seems wrong to me. But really the main issue for me is that this amount of wind should not move the aircraft or the prop. On 07/05/2012 04:14 PM, Hyde Yamakawa wrote: Does this happen only left wind, right? Or happens on right wind too? Hyde (2012年07月05日 10:02), Viktor Radnai wrote: Technically, you also have the compression of the wheels -- when the wheel turns, part of the tire in the front gets compressed while the back gets uncompressed. This is also a friction loss. Plus there is some friction between the rubber and the road. But my point is that the amount of friction is wrong. The plane should not start rolling with a 3 knot wind, that's nonsense. Static friction is greater than dynamic friction. If you stop your car on moderately flat terrain with no brakes, it will not start to roll easily, not even if it's windy. If the terrain is flat, it will stay put. If the slope is greater, it might start to roll very slowly with the bearings turning in stick/slip mode and then finally it will just roll and accelerate. So the plane should stay put until there's a 20-30 knot wind blowing against it. Probably more on grass. On grass with the real plane you might need half throttle or more to get it moving. The funny thing is that this part seems to be modelled correctly in Flightgear, so no idea what's wrong with the effect of the wind :) Same goes for the prop, the resistance the shut down engine offers is too small and so it's turned over way too easily. Turning the prop of the Falke at about 2/3 of its span requires about the same amount of force as lifting an object that weights 1-2 kg. This is a rough guess and the actual force varies during the compression cycle but you get the idea. For the wheels, I can try to make sure that the brakes are always slightly set, but what do I do for the prop? Cheers, Vik On 07/05/2012 03:30 PM, Emilian Huminiuc wrote: On Thursday 05 July 2012 15:21:24 Viktor Radnai wrote: 1. When the aircraft is parked with no parking brake, it will usually start to roll slowly backwards -- pushed by the wind and maybe the runway slope. If I start the engine on idle, the thrust generated by the idle prop might stop this roll. On tarmac, even a 3 knot wind is enough to start pushing the plane back. On grass more is needed -- maybe 20 knots? Hi, In that case, even in real life, there's no other friction at play than the friction inside the wheel bearings, friction which is very low, almost ignorable. Regards, Emilian -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim static friction
Hi, On Friday 22 May 2009 16:55:38 d...@bighost.com.br wrote: Understood, using the simulation time in the groundcache makes it much easier to track. Now it is just a matter of accounting for the FDM and the integrator iteration delays in relation to the cache_time_offset. Hopefully with you patch and this explanation I will be able to finish it next week. Thank you again. You are welcome. Tell me when you need anything. Greetings Mathias -- Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT is a gathering of tech-side developers brand creativity professionals. Meet the minds behind Google Creative Lab, Visual Complexity, Processing, iPhoneDevCamp asthey present alongside digital heavyweights like Barbarian Group, R/GA, Big Spaceship. http://www.creativitycat.com ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim static friction
Hi Diego,
On Tuesday 19 May 2009 03:32:53 Diogo Kastrup wrote:
I am sorry for the delay, finally I could get back to this. Besides
moving to a new city/job, I have also bought a model airplane, so I feel
like a child again with my new toy wasting all my free time playing.
Have fun :)
I am changing the friction code to use the new API, but I am having some
trouble. It looks like the velocities returned by the get_body method
are always zero. Maybe the problem is in the
void BodyFinder::apply(BVHMotionTransform transform)
method as even when it finds the right ID it returns before the
velocities are set.
I have not tested that part of the api. So testers and comments welcome.
Hmm, ok. I see.
Does the attached patch help?
With your code, you might have a better test bed than I have.
Also I am having some trouble to synchronize the stuck point with the
carrier movement again. I saw that there are some changes to the time
offset passed to the Ground Callback in YASim.cxx. What should I
expect to be already taken care by get_body? Is it compensating for the
delay between the AI update and the FDM iterations? And what about the
Integrator iterations? This was the hardest part of the friction patch
so any tip would be very helpful.
*In* *theory* the groundcache expects times that corespond to the simulation
time. FGInterface accounts for the difference between the start of the FDM
instance and the simulation time (that is the cache_time_offset in
FGInterface).
I believe that YASim did not track the simulation time itself, so I added
that? True? Also, YASim internal integration steps do not know about the exact
time they are for. You might need to take care of them.
There should be no delay between the FDM and AI iterations. The groundcache is
valid for a specific time interval. And when you start integration, you should
be at the start of that interval. Once you are ready for that timeslice, you
should be no further than the end of the caches validity.
This is not checked.
You may enable the ground cache debug code be defining the GROUNDCACHE_DEBUG in
groundcache.hxx. Then you can play with /fdm/groundcache-debug-level in the
property tree and see what the groundcache contains. That helped me to
understand the timeslicing in flightgear.
Greetings
Mathias
Index: groundcache.cxx
===
RCS file: /var/cvs/FlightGear-0.9/source/src/FDM/groundcache.cxx,v
retrieving revision 1.43
diff -u -r1.43 groundcache.cxx
--- groundcache.cxx 16 Mar 2009 09:48:18 - 1.43
+++ groundcache.cxx 22 May 2009 13:50:00 -
@@ -474,11 +474,10 @@
if (_id == transform.getId()) {
_foundId = true;
-return;
+} else {
+transform.traverse(*this);
}
-transform.traverse(*this);
-
if (_foundId) {
SGMatrixd toWorld = transform.getToWorldTransform(_time);
SGVec3d referencePoint = _bodyToWorld.xformPt(SGVec3d::zeros());
--
Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT
is a gathering of tech-side developers brand creativity professionals. Meet
the minds behind Google Creative Lab, Visual Complexity, Processing,
iPhoneDevCamp asthey present alongside digital heavyweights like Barbarian
Group, R/GA, Big Spaceship. http://www.creativitycat.com ___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim static friction
Hi Mathias, I have not tested that part of the api. So testers and comments welcome. Hmm, ok. I see. Does the attached patch help? With your code, you might have a better test bed than I have. Thank you for the patch, I will be out of town this weekend so I will try it next week. *In* *theory* the groundcache expects times that corespond to the simulation time. FGInterface accounts for the difference between the start of the FDM instance and the simulation time (that is the cache_time_offset in FGInterface). I believe that YASim did not track the simulation time itself, so I added that? True? Also, YASim internal integration steps do not know about the exact time they are for. You might need to take care of them. There should be no delay between the FDM and AI iterations. The groundcache is valid for a specific time interval. And when you start integration, you should be at the start of that interval. Once you are ready for that timeslice, you should be no further than the end of the caches validity. This is not checked. Understood, using the simulation time in the groundcache makes it much easier to track. Now it is just a matter of accounting for the FDM and the integrator iteration delays in relation to the cache_time_offset. Hopefully with you patch and this explanation I will be able to finish it next week. Thank you again. Regards, Diogo -- Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT is a gathering of tech-side developers brand creativity professionals. Meet the minds behind Google Creative Lab, Visual Complexity, Processing, iPhoneDevCamp asthey present alongside digital heavyweights like Barbarian Group, R/GA, Big Spaceship. http://www.creativitycat.com ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim static friction
I'm sorry for the distraction, but since you brought it up, you have to tell us about your model airplane! :-) Lol, of course, I should have imagined :) There is a small lake near my house I can walk to, so I have been having a blast on calm evenings with my Seawind park flyer: http://baron.flightgear.org/~curt/Models/Current/SeaWindEP/ I also got my Shrike 40 flying recently ... that is a real kick ... bank and yank style flying, but it slows way up for pretty tame landings. (It's about the world's simplest ARF so it only took me 5 months to assemble): http://baron.flightgear.org/~curt/Models/Current/Shrike40/ I just ordered a Polaris park flyer kit ... usually takes me at least a year to put together an ARF and this is a kit, so no promises when this will be ready to fly, but it should be nice at the little lake near my house... http://www.rcgroups.com/forums/showthread.php?t=954314 There are two HD (95Mb) movies you can download from the rcgroups thread that are probably worth the bandwidth if you enjoy cool airplanes. Those are some really nice models. Mine is actually just simple trainer called Ready Mk3. It is made of ABS plastic and theoretically it is very resistant to impacts (what is important as it is my first model airplane). The place where I fly it has some lakes and there is a floats kit for this plane, so I hope I will do some water landings eventually. I have no pictures of it on the net yet, but you can see it on the manufacturer's website: http://www.thundertiger.com/product/4591-K.html Regards, Diogo -- Crystal Reports - New Free Runtime and 30 Day Trial Check out the new simplified licensing option that enables unlimited royalty-free distribution of the report engine for externally facing server and web deployment. http://p.sf.net/sfu/businessobjects ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim static friction
Mathias Fröhlich escreveu: Prototyped and checked in. I don't claim that it is bug free. So help chasing them :) Feel free to ask if something is too bad documented or does not work as expected. Hi Mathias, I am sorry for the delay, finally I could get back to this. Besides moving to a new city/job, I have also bought a model airplane, so I feel like a child again with my new toy wasting all my free time playing. I am changing the friction code to use the new API, but I am having some trouble. It looks like the velocities returned by the get_body method are always zero. Maybe the problem is in the void BodyFinder::apply(BVHMotionTransform transform) method as even when it finds the right ID it returns before the velocities are set. Also I am having some trouble to synchronize the stuck point with the carrier movement again. I saw that there are some changes to the time offset passed to the Ground Callback in YASim.cxx. What should I expect to be already taken care by get_body? Is it compensating for the delay between the AI update and the FDM iterations? And what about the Integrator iterations? This was the hardest part of the friction patch so any tip would be very helpful. Regards, Diogo -- Crystal Reports - New Free Runtime and 30 Day Trial Check out the new simplified licensing option that enables unlimited royalty-free distribution of the report engine for externally facing server and web deployment. http://p.sf.net/sfu/businessobjects ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim static friction
Hi, On Thursday 05 March 2009 22:08:54 Diogo Kastrup wrote: That sounds great, I am looking forward to using this improved groundcache. I couldn't test it so far but I will try to update the friction code ASAP. Unfortunately I am going to stay away from fgfs a little longer than expected. What is the status of the the body fixed reference point API? I would be glad to help on you with that when I get back to it. Prototyped and checked in. I don't claim that it is bug free. So help chasing them :) Feel free to ask if something is too bad documented or does not work as expected. Greetings Mathias -- Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise -Strategies to boost innovation and cut costs with open source participation -Receive a $600 discount off the registration fee with the source code: SFAD http://p.sf.net/sfu/XcvMzF8H ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim static friction
Hello, Mathias Fröhlich wrote: Ok, I am working on a major overhaul to speed up and generalize the ground intersection stuff. So parts of what you code is already covered by that more general approach. When I look at your patch, the only thing that is missing so far in the ground intersection api is the body fixed reference point of the body you are rolling on. I would suggest that we add this kind of reference point to the existing api so that you can make use of that. As I told in an other mail, the way the surface velocities are computed and provided to the FDM have subtle problems in the current implementation. May be this will already improve the way you stick on moving surfaces. That sounds great, I am looking forward to using this improved groundcache. I couldn't test it so far but I will try to update the friction code ASAP. Unfortunately I am going to stay away from fgfs a little longer than expected. What is the status of the the body fixed reference point API? I would be glad to help on you with that when I get back to it. Regards, Diogo -- Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise -Strategies to boost innovation and cut costs with open source participation -Receive a $600 discount off the registration fee with the source code: SFAD http://p.sf.net/sfu/XcvMzF8H ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim static friction
Hi Diego, On Thursday 05 March 2009 22:08:54 Diogo Kastrup wrote: That sounds great, I am looking forward to using this improved groundcache. I couldn't test it so far but I will try to update the friction code ASAP. Unfortunately I am going to stay away from fgfs a little longer than expected. What is the status of the the body fixed reference point API? I would be glad to help on you with that when I get back to it. Well, currently working on two problems. Still the memory stuff, which is, solved right, a tricky thing. And your interface as well as other useful interface additions. What I can probably provide this weekend is something similar in the api to your current additions without the need to configure extra 'platforms'. I am not sure if this is the right approach, but It will provide you with the information you need to have a reference position/orientation. So, in principle: Each get_agl call delivers additionally an opaque id of the body you have found in the agl computation. This id might be 0 which means you are standing on something earth fixed. If you get a nonzero id value, you can with a new call 'get_body' get the transform matrix and velocity of that body with a given id and at a given time. This works as long as you have anything from this body in the groundcache which should be true for your implemented 5mm spring. So the difference api wise is that you need to track the 'platform id' yourself instead of doing this in the groundcache/interface. Think of the case where the aircraft stands with the nose wheel on the carriers main deck and with the main wheels on an elevator. You will recieve the carriers's main id for the nose wheel and the elevator id for the main wheels. If you move the elevator down a bit, this still works. Does this sound reasonable for you? Additionally, there will be queries for 'nearest geometry in any direction closer than a given maximum distance'. Remember that the agl thing evaluates the nearest geometry looking downwards without distance limitation. Usually this type of environment query is a little cheaper since the geometry what is queried is a small sphere instead of a large line, so less geometry is tested. You will also get the body id here ... This could be helpful for simple contact points. Greetings Mathias -- Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise -Strategies to boost innovation and cut costs with open source participation -Receive a $600 discount off the registration fee with the source code: SFAD http://p.sf.net/sfu/XcvMzF8H ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim static friction
Hi, On Monday 23 February 2009 03:41:17 Diogo Kastrup wrote: I finally got this working with the carrier, here is the patch. Please let me know what you guys think about it. Features: - No more sliding when should be stopped. - better behavior when cornering - better braking behavior * Problems: - More information about the brake system is needed to simulate the wheel locks better. This requires changes to the aircraft definition format. For now we may need to slow down the reaction to the brake button so we can stop planes easier without sliding. Brakes mapped to an axis should work much better. - More information about the tire in general could help improving the simulation. I may be a little busy and offline this week as I am moving to another city, but if you guys decide to commit this patch and some changes are required, just let me know and I will work on it ASAP. Ok, I am working on a major overhaul to speed up and generalize the ground intersection stuff. So parts of what you code is already covered by that more general approach. When I look at your patch, the only thing that is missing so far in the ground intersection api is the body fixed reference point of the body you are rolling on. I would suggest that we add this kind of reference point to the existing api so that you can make use of that. As I told in an other mail, the way the surface velocities are computed and provided to the FDM have subtle problems in the current implementation. May be this will already improve the way you stick on moving surfaces. Greetings Mathias -- Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise -Strategies to boost innovation and cut costs with open source participation -Receive a $600 discount off the registration fee with the source code: SFAD http://p.sf.net/sfu/XcvMzF8H ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim static friction
Csaba, thank you very much for testing. See my comments below. Em Seg, 2009-02-23 às 04:59 +0100, Csaba Halász escreveu: I tested with the bo105 on solid ground first. It doesn't move an inch with engines shut down. During startup, it doesn't turn either. The s76c (which has wheels instead of skids) vibrates during startup (we can call that realistic ;)) That is weird, I will try to find out why. After a quick look, I am concerned about the values of the spring and compression in the definition file of the s76c. The static friction value is very high also. I will do some tests... (...) Apparently it happens because I have the carrier stationary. Setting it into motion, the bo105 starts up fine. With the s76c the carrier sails out from underneath even with the parking brake on (which is the default). With the v22, it only happens with the brakes off, which seems right. Setting the wind so that it cancels out the relative wind due to motion makes the v22 stay in place even with brakes off. Haven't tried maneuvering the Nimitz yet. The s76c problem might be an aircraft specific issue and the zero-speed carrier error sounds easy to fix. All in all it looks quite good so far. How can I set the carrier speed to zero? :) I would like to reproduce it here. Anyway, I think that moving the following lines to inside the if (_slipping) loop around line 650 in Gear.cpp solves it: // Get the direction of movement float dir[3]; Math::unit3(cv, dir); The carrier sailing out from underneath is strange, it looks like the groundcache can't finding a intersection with the ground for these aircrafts and it is using a fallback value. Can someone check if this happens without my patch also? I will try to find out why this is happening. Regards, Diogo -- Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise -Strategies to boost innovation and cut costs with open source participation -Receive a $600 discount off the registration fee with the source code: SFAD http://p.sf.net/sfu/XcvMzF8H ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASim static friction
Em Seg, 2009-02-23 às 17:58 -0300, Diogo Kastrup escreveu:
The carrier sailing out from underneath is strange, it looks like the
groundcache can't finding a intersection with the ground for these
aircrafts and it is using a fallback value. Can someone check if this
happens without my patch also? I will try to find out why this is
happening.
Yep, the groundcache can't find a intersection. I think this problem
occurs without my patch also. To fix it just change groundcache.cxx
around line 312 from this:
if (d*d reference_vehicle_radius*dot(n, n)) {
to this:
if (d*d reference_vehicle_radius*reference_vehicle_radius*dot(n, n)) {
I am sending the diff again with the fix for this and the (untested) fix
for the zero-speed problem, just in case someone want to play with it.
I am still trying to figure out why the s76c is shaking. This happens
with other helicopters also, like the R22.
Regards,
Diogo
Index: src/FDM/flight.cxx
===
RCS file: /var/cvs/FlightGear-0.9/source/src/FDM/flight.cxx,v
retrieving revision 1.35
diff -u -r1.35 flight.cxx
--- src/FDM/flight.cxx 9 Nov 2007 05:39:13 - 1.35
+++ src/FDM/flight.cxx 23 Feb 2009 23:14:47 -
@@ -705,6 +705,15 @@
return dist*SG_METER_TO_FEET;
}
+void
+FGInterface::get_platform(double t, double pos[3], float *orient)
+{
+ SGVec3d _pos;
+ ground_cache.get_platform(t, _pos, orient);
+ if (pos)
+assign(pos, _pos);
+}
+
// Legacy interface just kept because of JSBSim
bool
FGInterface::get_agl_m(double t, const double pt[3],
Index: src/FDM/flight.hxx
===
RCS file: /var/cvs/FlightGear-0.9/source/src/FDM/flight.hxx,v
retrieving revision 1.18
diff -u -r1.18 flight.hxx
--- src/FDM/flight.hxx 27 Jul 2008 16:25:14 - 1.18
+++ src/FDM/flight.hxx 23 Feb 2009 23:14:48 -
@@ -629,6 +629,9 @@
double end[2][3], double vel[2][3]);
double get_cat_ft(double t, const double pt[3],
double end[2][3], double vel[2][3]);
+
+// Return the position and orientation of the platform
+void get_platform(double t, double pos[3], float *orient);
// Return the altitude above ground below the wgs84 point pt
@@ -642,8 +645,8 @@
int *type, double *loadCapacity,
double *frictionFactor, double *agl);
bool get_agl_m(double t, const double pt[3],
- double contact[3], double normal[3], double vel[3],
- int *type, const SGMaterial **material,double *agl);
+ double contact[3], double normal[3], double vel[3],
+ int *type, const SGMaterial **material,double *agl);
bool get_agl_ft(double t, const double pt[3],
double contact[3], double normal[3], double vel[3],
int *type, double *loadCapacity,
Index: src/FDM/groundcache.cxx
===
RCS file: /var/cvs/FlightGear-0.9/source/src/FDM/groundcache.cxx,v
retrieving revision 1.33
diff -u -r1.33 groundcache.cxx
--- src/FDM/groundcache.cxx 15 Feb 2009 00:56:22 - 1.33
+++ src/FDM/groundcache.cxx 23 Feb 2009 23:14:49 -
@@ -158,6 +158,7 @@
_material(0),
cache_ref_time(0.0),
wire_id(0),
+ platform_id(0),
reference_wgs84_point(SGVec3d(0, 0, 0)),
reference_vehicle_radius(0.0),
down(0.0, 0.0, 0.0),
@@ -198,6 +199,8 @@
gp.vel = SGVec3d(0.0, 0.0, 0.0);
gp.rot = SGVec3d(0.0, 0.0, 0.0);
gp.pivot = SGVec3d(0.0, 0.0, 0.0);
+gp.platform_id = 0;
+gp.platform = NULL;
gp.material = 0;
backfaceCulling = false;
// XXX state set might be higher up in scene graph
@@ -235,6 +238,8 @@
break;
default:
gp.type = FGInterface::Solid;
+gp.platform_id = ud-carrier-getID();
+gp.platform = ud-carrier;
break;
}
// Copy the velocity from the carrier class.
@@ -304,7 +309,7 @@
// Only check if the triangle is in the cache sphere if the plane
// containing the triangle is near enough
double d = dot(n, v[0] - localCacheReference);
-if (d*d reference_vehicle_radius*dot(n, n)) {
+if (d*d reference_vehicle_radius*reference_vehicle_radius*dot(n, n)) {
// Check if the sphere around the vehicle intersects the sphere
// around that triangle. If so, put that triangle into the cache.
double r2 = boundRadius + reference_vehicle_radius;
@@ -323,6 +328,10 @@
t.sphere.setRadius(boundRadius);
t.gp = gp;
triangles.push_back(t);
+if (gp.platform_id
+ (platforms.empty()
+|| dynamic_castclass FGAIBase*(platforms.back())-getID() != gp.platform_id))
+platforms.push_back(gp.platform);
}
