>-- Messaggio originale --
>Date: Thu, 29 Apr 2004 10:19:08 -0700
>From: Andy Ross <[EMAIL PROTECTED]>
>To: FlightGear developers discussions <[EMAIL PROTECTED]>
>Subject: Re: [Flightgear-devel] YASIM question (for Andy Ross)
>Reply-To: FlightGear developers discussions <[EMAIL PROTECTED]>
>Marco Gugel wrote:
>> My idea is to develop a truck simulation inside FlightGear and I
>> have thought to start from Yasim because it uses the rigidbody
>Right. That's the RigidBody/Integrator/BodyEnvironment
>implementation. You set masses on the body, calculate forces inside
>the environment, and let the integrator figure out the result.
>> But there is a "little" problem: the collision detection, which is
>> not implemented. Moreover the airplanes on the gorund don't follow
>> the slope of the terrain in a realistic manner: [...] is this
>> behavior due to the fact that the only point of reference is the cg
>> of the aircraft and not the position of the wheels?
>This is a different problem, and not the only one you are going to
>discover. Indeed, the collision detection is a hack right now, which
>presumes a horizontal ground plane under the aircraft. This works
>just fine for airfields, but not so well for hills (or ships/carriers,
>for that matter).
>It's not related to c.g. or positioning issues at all; in fact the
>application of force at the position of the wheels *is* modelled
>correctly in YASim, and the c.g. is computed dynamically from the mass
>distribution; it isn't a "user visible" parameter... All (heh) that
>is required to fix this bug is for the gear model to calculate a
>proper intersection with the terrain polygons instead of using a
>ground plane defined for the whole aircraft.
Is this solution reasonable to implement or is it too difficult? It's only
a week that I study FlightGear code and now I don't know if this modify
is easy or not to do?
>A more serious problem, though, is that the current YASim gear force
>model works rather poorly for ground vehicles. It "slips" against
>side forces instead of holding position. See recent posts about the
>gear model for more notes. A few ideas have been kicked around, but
>all of them are kinda scary to implement.
For the gear model I think I'll use a different wheel model.
However, the main problem remain the collision detection (I have doubts
that it could be done on flightgear): my idea is to identify collision between
the truck and other aircraft/buildings on the scene, but still I don't understand
how in the code the single object of the scene are loaded!! I know that
the scenery is loaded in tiles dinamically using a cache method, but I have
to investigate more about!
Flightgear-devel mailing list