* Topic for #brlcad is: http://brlcad.org/ || BRL-CAD is now Free 
Software! || 7.6.6 to be released by the 15th!
* Topic for #brlcad set by learner at Wed Dec 14 00:12:51 2005
fenn ahoy mateys!
fenn i'm looking for a relatively simple solid modeling language to 
simulate collisions in machine tools
fenn stuff like cylinder, cube, sphere, cone, prism
fenn also testing for collisions with a nurbs toolpath would be nice
fenn what parts of the brlcad tree would i need for this? or is brlcad the 
wrong tool for the job?
brlcad fenn, brl-cad wouldn't be the wrong tool for the job, but depending 
on exactly what you want changes what you might want to look at
brlcad and coding will probably be required
fenn yeah definitely
brlcad librt is where all the primitive and geometry introspect (via ray 
shooting) lives
fenn what handles geometry intersections?
fenn oh the other tricky thing is i have to transform some things into n 
dimensional space
brlcad it's implicit geometry representations
brlcad so geometry intersections are known when a ray is fired
brlcad and the implicit equations and segment/partition lists are computed
brlcad you can register callback handles via the raytrace programming api 
for hit/miss/overlap
fenn so a ray is like a vector? it has magnitude?
brlcad a ray has a starting position and a direction
fenn hmmm
brlcad you can get geometry information along that ray
fenn i need to know if a vector is inside a volume
fenn brlcad is probably the wrong tool for this
brlcad if your talking about what I think you're talking about
brlcad then you have some "tool", a shaped carving/cutting bit with some 
3d shape
fenn eh, no that's a different project
brlcad oh?  then what the "testing for collsions with a nurbs toolpath" 
mean?
fenn ok these projects are all related you can see
fenn i have a CAM project that takes a solid and plans out the toolpaths
brlcad k
fenn now we are working on a new CNC motion controller to replace emc
fenn they have some similar functions and some redundant functions
fenn like the cnc controller has to be able to compensate for cutters of 
different sizes
fenn so if you have are cutting a sharp inside corner for instance, you 
dont want to blindly offset the path by some amount
fenn this kinda hard without pictures
brlcad i think i follow though
fenn so, cutter comp and toolpath planning in the CAM are pretty similar
brlcad you want a sharp corner, and if you were using a large circular 
cut, you'd have a big corner curve
fenn yeah
fenn if you follow the path that was made for a smaller cutter it might go 
too far into the corner and gouge the part
fenn so i wanted to do collision detection between the toolpath and the 
cutter
fenn make sense?
brlcad now do you restrict the toolpath to "mostly" 2D (like a thin slab 
of metal being cut), or entirely 2D (sheet-metal style cutting), or 
entirely 3d?
fenn it's restricted to mostly 2d 
fenn i think it's kinda dumb myself
fenn why not just run it thru cam again?
fenn but anyway..
brlcad so basically it's a 2D cutting path and toolbit exercise where you 
just have to take the vertical shape of the toolbit into consideration
fenn the main problem i've been thinking about is how to transform 
arbitrary geometries from "cartesian space" to "joint space"
fenn where "joint space" is the space with a set of basis vectors defined 
by the actual motions of the machine tool
fenn like a milling machine will have a quill, a knee, an X axis and a Y 
axis
fenn so in this instance you are taking a 3d toolpath and transforming it 
into 4 dimensional space
fenn now you want to make sure that you are within a specified tolerance, 
but the information in realtime is coming from the joints
brlcad have you run rtedge in brl-cad yet?
fenn so i think it's easier to transform the tolerance into joint space 
before the machine moves than to try to transform all the joints back into 
normal cartesian coordinates
fenn no.. *hangs his head*
fenn i get lost every time i try to use brlcad
brlcad that effectively solves the path contour, detecting/reporting the 
edge of a specified slice of geometry
brlcad you could even use the results as basis vectors for a 2D bezier 
path directly
fenn like slicing a part along the z axis?
brlcad you'd have to account for the direction changes to detect 
inside/outside corners that would need a different toolbit
brlcad again, as you noted, pictures would help :)
fenn i figured this is the sort of stuff you did in simulations
* Maloeran has quit (Read error: 110 (Connection timed out))
brlcad it is in many ways
brlcad those are almost always custom algorithms but the basic need is for 
geometric introspection, which is what librt provides
* Maloeran (n=alexis@modemcable065.3-83-70.mc.videotron.ca) has joined 
#brlcad
brlcad full blown simulations of a munition hitting a target can be 
simulated that way
fenn is there a way to get librt without all the rest of brlcad?
brlcad or energy/blast effects, etc
brlcad heh
fenn if not it's not a big deal
brlcad you could, sure .. but "why"? :)
fenn for instance if you're running a cnc controller off a usb stick and 
dont have space for all of brlcad
brlcad ah, for that purpose, sure
fenn cool.. i'll have to look into it more then
brlcad librt relies upon brl-cad's basic numerics and utility libraries 
(libbn and libbu), libregex, and libz 
brlcad so you pull those four directories out, and you'd have just what 
you needed, a fairly contained library
fenn they're just shared library files right? (.so)
brlcad I could make a configure target for that as well for convenience
brlcad shared or static
fenn well, .a, close enough
fenn i can just link to that at compile time