[ODE] bouncy car

[Luke Kenneth Casson Leighton]

On Mon, Nov 08, 2004 at 11:50:24AM -0700, Megan Fox wrote:
> It is not uncommon for people to model their cars as single boxes, not
> even using true wheels (cast rays straight down at each of the four
> corners, roll your own springy suspension code, place the tires
> accordingly along the ray, etc), depending on the number of cars
> you'll want in a sim and the sort of handling you want.
> 
> Especially in a case like that, how you're going to go about applying
> your wind resistance/downforce/upforce/etc is limited only by your own
> imagination.  From previous discussions, people have found applying a
> down force to greatly help with stability (with some kooky black magic
> involving how many tires are touching ground to make sure you're not
> accelerating your car into space with this down force when it's in the
> process of rolling), as well as making sure to place your center of
> mass at the lower edge of the car's box body (that is, translate your
> box geom up the Y axis by boxheight/2 relative to the center of the
> car, via an nGeomTransform).

 dear megan,

 your advice about going over the archives is greatly appreciated: i do
 find, however, that actually _finding_ anything in mailing list
 archives is a tricky hit-and-miss affair.

 if anyone on this list with a working history can remember some
 references, keywords, rough dates, subject lines, people in the
 discussions, i would be _really_ grateful.

 one thing i tend to find missing from pretty much every mailing list i
 come across is a FAQ referring to the archives, e.g. in this case:

 Q: what discussions have there been involving car designs?
 A: http://archives.q12.org/blah was a good one on wind resistance
    http.... was a good one on suspension

 etc.

 anyway.

 what i have managed so far works pretty successfully, acting pretty
 much like a high performance vehicle - just like people do with stolen
 vehicles doing donuts in car parks, that sort of thing.

 the one behaviour i _did_ find slightly odd is "crabbing" - going
 sideways from a standing start with full steering lock to the left:
 that's nothing i've ever seen anyone make a _real_ car do!

 the behavioural bugs i am seeing are:

 - front wheels go haywire in all directions.  i reduced "fudge factor"
   on the Hinge2 front joints to 0.0001 and it seemed to disappear.

 - at high speed, the front steering doesn't work - steering left or
   right results in no movement (!!!)

 i've yet to try StopCRM and will let you know how it goes.

 l.